Compare commits
9 Commits
4bit-optim
...
main-base
| Author | SHA1 | Date | |
|---|---|---|---|
|
ff939d8a64 | ||
|
324d59ea0d | ||
|
|
f1ebaa07c6 | ||
|
34ba634b8c | ||
|
4e69aa48ab | ||
|
|
629450cecd | ||
|
|
2a1589f6f6 | ||
|
7d55607368 | ||
|
|
7803f0934f |
31
.github/workflows/docs.yml
vendored
Normal file
31
.github/workflows/docs.yml
vendored
Normal file
@@ -0,0 +1,31 @@
|
||||
name: Publish Docs
|
||||
on:
|
||||
push:
|
||||
branches:
|
||||
- main
|
||||
|
||||
permissions:
|
||||
contents: write
|
||||
pages: write
|
||||
|
||||
jobs:
|
||||
build-deploy:
|
||||
runs-on: ubuntu-latest
|
||||
steps:
|
||||
- name: Check out repository
|
||||
uses: actions/checkout@v4
|
||||
- name: Set up Quarto
|
||||
uses: quarto-dev/quarto-actions/setup@v2
|
||||
- name: Setup Python
|
||||
uses: actions/setup-python@v3
|
||||
with:
|
||||
python-version: '3.10'
|
||||
- name: install dependencies
|
||||
run: |
|
||||
python3 -m pip install jupyter
|
||||
- name: Publish to GitHub Pages (and render)
|
||||
uses: quarto-dev/quarto-actions/publish@v2
|
||||
with:
|
||||
target: gh-pages
|
||||
env:
|
||||
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
|
||||
2
.github/workflows/tests.yml
vendored
2
.github/workflows/tests.yml
vendored
@@ -34,7 +34,7 @@ jobs:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
python_version: ["3.10", "3.11"]
|
||||
timeout-minutes: 10
|
||||
timeout-minutes: 20
|
||||
|
||||
steps:
|
||||
- name: Check out repository code
|
||||
|
||||
3
.gitignore
vendored
3
.gitignore
vendored
@@ -2,6 +2,7 @@
|
||||
configs
|
||||
last_run_prepared/
|
||||
.vscode
|
||||
_site/
|
||||
|
||||
# Byte-compiled / optimized / DLL files
|
||||
__pycache__/
|
||||
@@ -172,3 +173,5 @@ wandb
|
||||
lora-out/*
|
||||
qlora-out/*
|
||||
mlruns/*
|
||||
|
||||
/.quarto/
|
||||
|
||||
23
README.md
23
README.md
@@ -32,6 +32,7 @@ Features:
|
||||
- [Bare Metal Cloud GPU](#bare-metal-cloud-gpu)
|
||||
- [Windows](#windows)
|
||||
- [Mac](#mac)
|
||||
- [Google Colab](#google-colab)
|
||||
- [Launching on public clouds via SkyPilot](#launching-on-public-clouds-via-skypilot)
|
||||
- [Dataset](#dataset)
|
||||
- [How to Add Custom Prompts](#how-to-add-custom-prompts)
|
||||
@@ -42,8 +43,8 @@ Features:
|
||||
- [Merge LORA to Base](#merge-lora-to-base)
|
||||
- [Special Tokens](#special-tokens)
|
||||
- Advanced Topics
|
||||
- [Multipack](./docs/multipack.md)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
|
||||
- [RLHF & DPO](./docs/rlhf.md)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
|
||||
- [Multipack](./docs/multipack.qmd)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
|
||||
- [RLHF & DPO](./docs/rlhf.qmd)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
|
||||
- [Common Errors](#common-errors-)
|
||||
- [Tokenization Mismatch b/w Training & Inference](#tokenization-mismatch-bw-inference--training)
|
||||
- [Debugging Axolotl](#debugging-axolotl)
|
||||
@@ -149,7 +150,7 @@ accelerate launch -m axolotl.cli.train https://raw.githubusercontent.com/OpenAcc
|
||||
```
|
||||
|
||||
>[!Tip]
|
||||
> If you want to debug axolotl or prefer to use Docker as your development environment, see the [debugging guide's section on Docker](docs/debugging.md#debugging-with-docker).
|
||||
> If you want to debug axolotl or prefer to use Docker as your development environment, see the [debugging guide's section on Docker](docs/debugging.qmd#debugging-with-docker).
|
||||
|
||||
<details>
|
||||
|
||||
@@ -267,7 +268,11 @@ Use the below instead of the install method in QuickStart.
|
||||
```
|
||||
pip3 install -e '.'
|
||||
```
|
||||
More info: [mac.md](/docs/mac.md)
|
||||
More info: [mac.md](/docs/mac.qmd)
|
||||
|
||||
#### Google Colab
|
||||
|
||||
Please use this example [notebook](examples/colab-notebooks/colab-axolotl-example.ipynb).
|
||||
|
||||
#### Launching on public clouds via SkyPilot
|
||||
To launch on GPU instances (both on-demand and spot instances) on 7+ clouds (GCP, AWS, Azure, OCI, and more), you can use [SkyPilot](https://skypilot.readthedocs.io/en/latest/index.html):
|
||||
@@ -409,7 +414,7 @@ pretraining_dataset: # hf path only
|
||||
{"segments": [{"label": true|false, "text": "..."}]}
|
||||
```
|
||||
|
||||
This is a special format that allows you to construct prompts without using templates. This is for advanced users who want more freedom with prompt construction. See [these docs](docs/input_output.md) for more details.
|
||||
This is a special format that allows you to construct prompts without using templates. This is for advanced users who want more freedom with prompt construction. See [these docs](docs/input_output.qmd) for more details.
|
||||
|
||||
##### Conversation
|
||||
|
||||
@@ -1125,7 +1130,7 @@ fsdp_config:
|
||||
|
||||
##### FSDP + QLoRA
|
||||
|
||||
Axolotl supports training with FSDP and QLoRA, see [these docs](docs/fsdp_qlora.md) for more information.
|
||||
Axolotl supports training with FSDP and QLoRA, see [these docs](docs/fsdp_qlora.qmd) for more information.
|
||||
|
||||
##### Weights & Biases Logging
|
||||
|
||||
@@ -1204,7 +1209,7 @@ although this will be very slow, and using the config options above are recommen
|
||||
|
||||
## Common Errors 🧰
|
||||
|
||||
See also the [FAQ's](./docs/faq.md) and [debugging guide](docs/debugging.md).
|
||||
See also the [FAQ's](./docs/faq.qmd) and [debugging guide](docs/debugging.qmd).
|
||||
|
||||
> If you encounter a 'Cuda out of memory' error, it means your GPU ran out of memory during the training process. Here's how to resolve it:
|
||||
|
||||
@@ -1238,7 +1243,7 @@ It's safe to ignore it.
|
||||
|
||||
> NCCL Timeouts during training
|
||||
|
||||
See the [NCCL](docs/nccl.md) guide.
|
||||
See the [NCCL](docs/nccl.qmd) guide.
|
||||
|
||||
|
||||
### Tokenization Mismatch b/w Inference & Training
|
||||
@@ -1256,7 +1261,7 @@ Having misalignment between your prompts during training and inference can cause
|
||||
|
||||
## Debugging Axolotl
|
||||
|
||||
See [this debugging guide](docs/debugging.md) for tips on debugging Axolotl, along with an example configuration for debugging with VSCode.
|
||||
See [this debugging guide](docs/debugging.qmd) for tips on debugging Axolotl, along with an example configuration for debugging with VSCode.
|
||||
|
||||
## Need help? 🙋
|
||||
|
||||
|
||||
51
_quarto.yml
Normal file
51
_quarto.yml
Normal file
@@ -0,0 +1,51 @@
|
||||
project:
|
||||
type: website
|
||||
|
||||
website:
|
||||
title: "Axolotl"
|
||||
description: "Fine-tuning"
|
||||
favicon: favicon.jpg
|
||||
navbar:
|
||||
title: Axolotl
|
||||
background: dark
|
||||
pinned: false
|
||||
collapse: false
|
||||
tools:
|
||||
- icon: twitter
|
||||
href: https://twitter.com/axolotl_ai
|
||||
- icon: github
|
||||
href: https://github.com/OpenAccess-AI-Collective/axolotl/
|
||||
- icon: discord
|
||||
href: https://discord.gg/7m9sfhzaf3
|
||||
|
||||
sidebar:
|
||||
pinned: true
|
||||
collapse-level: 2
|
||||
style: docked
|
||||
contents:
|
||||
- text: Home
|
||||
href: index.qmd
|
||||
- section: "How-To Guides"
|
||||
contents:
|
||||
# TODO Edit folder structure after we have more docs.
|
||||
- docs/debugging.qmd
|
||||
- docs/multipack.qmd
|
||||
- docs/fdsp_qlora.qmd
|
||||
- docs/input_output.qmd
|
||||
- docs/rlhf.qmd
|
||||
- docs/nccl.qmd
|
||||
- docs/mac.qmd
|
||||
- docs/multi-node.qmd
|
||||
- section: "Reference"
|
||||
contents:
|
||||
- docs/config.qmd
|
||||
- docs/faq.qmd
|
||||
|
||||
|
||||
|
||||
|
||||
format:
|
||||
html:
|
||||
theme: materia
|
||||
css: styles.css
|
||||
toc: true
|
||||
@@ -1 +1 @@
|
||||
This directory contains example config files that might be useful for debugging. Please see [docs/debugging.md](../docs/debugging.md) for more information.
|
||||
This directory contains example config files that might be useful for debugging. Please see [docs/debugging.qmd](../docs/debugging.qmd) for more information.
|
||||
|
||||
2
docs/.gitignore
vendored
Normal file
2
docs/.gitignore
vendored
Normal file
@@ -0,0 +1,2 @@
|
||||
/.quarto/
|
||||
_site/
|
||||
17
docs/config.qmd
Normal file
17
docs/config.qmd
Normal file
@@ -0,0 +1,17 @@
|
||||
---
|
||||
title: Config options
|
||||
description: A complete list of all configuration options.
|
||||
---
|
||||
|
||||
```{python}
|
||||
#|echo: false
|
||||
#|output: asis
|
||||
import re
|
||||
# Regex pattern to match the YAML block including its code fence
|
||||
pattern = r'<details[^>]*id="all-yaml-options"[^>]*>.*?<summary>All yaml options.*?```yaml(.*?)```.*?</details>'
|
||||
|
||||
with open('../README.md', 'r') as f:
|
||||
doc = f.read()
|
||||
match = re.search(pattern, doc, re.DOTALL)
|
||||
print("```yaml", match.group(1).strip(), "```", sep="\n")
|
||||
```
|
||||
@@ -1,4 +1,8 @@
|
||||
# Debugging Axolotl
|
||||
---
|
||||
title: Debugging
|
||||
description: How to debug Axolotl
|
||||
---
|
||||
|
||||
|
||||
This document provides some tips and tricks for debugging Axolotl. It also provides an example configuration for debugging with VSCode. A good debugging setup is essential to understanding how Axolotl code works behind the scenes.
|
||||
|
||||
18
docs/faq.md
18
docs/faq.md
@@ -1,18 +0,0 @@
|
||||
# Axolotl FAQ's
|
||||
|
||||
|
||||
> The trainer stopped and hasn't progressed in several minutes.
|
||||
|
||||
Usually an issue with the GPU's communicating with each other. See the [NCCL doc](../docs/nccl.md)
|
||||
|
||||
> Exitcode -9
|
||||
|
||||
This usually happens when you run out of system RAM.
|
||||
|
||||
> Exitcode -7 while using deepspeed
|
||||
|
||||
Try upgrading deepspeed w: `pip install -U deepspeed`
|
||||
|
||||
> AttributeError: 'DummyOptim' object has no attribute 'step'
|
||||
|
||||
You may be using deepspeed with single gpu. Please don't set `deepspeed:` in yaml or cli.
|
||||
21
docs/faq.qmd
Normal file
21
docs/faq.qmd
Normal file
@@ -0,0 +1,21 @@
|
||||
---
|
||||
title: FAQ
|
||||
description: Frequently asked questions
|
||||
---
|
||||
|
||||
|
||||
**Q: The trainer stopped and hasn't progressed in several minutes.**
|
||||
|
||||
> A: Usually an issue with the GPUs communicating with each other. See the [NCCL doc](nccl.qmd)
|
||||
|
||||
**Q: Exitcode -9**
|
||||
|
||||
> A: This usually happens when you run out of system RAM.
|
||||
|
||||
**Q: Exitcode -7 while using deepspeed**
|
||||
|
||||
> A: Try upgrading deepspeed w: `pip install -U deepspeed`
|
||||
|
||||
**Q: AttributeError: 'DummyOptim' object has no attribute 'step'**
|
||||
|
||||
> A: You may be using deepspeed with single gpu. Please don't set `deepspeed:` in yaml or cli.
|
||||
@@ -1,4 +1,10 @@
|
||||
# FDSP + QLoRA
|
||||
---
|
||||
title: FDSP + QLoRA
|
||||
description: Use FSDP with QLoRA to fine-tune large LLMs on consumer GPUs.
|
||||
format:
|
||||
html:
|
||||
toc: true
|
||||
---
|
||||
|
||||
## Background
|
||||
|
||||
@@ -1,4 +1,7 @@
|
||||
# Template-free prompt construction with the `input_output` format
|
||||
---
|
||||
title: Template-free prompt construction
|
||||
description: "Template-free prompt construction with the `input_output` format"
|
||||
---
|
||||
|
||||
<!-- TOC -->
|
||||
|
||||
@@ -1,8 +1,12 @@
|
||||
# Mac M series support
|
||||
---
|
||||
title: Mac M-series
|
||||
description: Mac M-series support
|
||||
---
|
||||
|
||||
Currently Axolotl on Mac is partially usable, many of the dependencies of Axolotl including Pytorch do not support MPS or have incomplete support.
|
||||
|
||||
Current support:
|
||||
|
||||
- [x] Support for all models
|
||||
- [x] Full training of models
|
||||
- [x] LoRA training
|
||||
@@ -1,4 +1,7 @@
|
||||
# Multi Node
|
||||
---
|
||||
title: Multi Node
|
||||
description: How to use Axolotl on multiple machines
|
||||
---
|
||||
|
||||
You will need to create a configuration for accelerate, either by using `accelerate config` and follow the instructions or you can use one of the preset below:
|
||||
|
||||
@@ -1,4 +1,7 @@
|
||||
# Multipack (Sample Packing)
|
||||
---
|
||||
title: Multipack (Sample Packing)
|
||||
description: Multipack is a technique to pack multiple sequences into a single batch to increase training throughput.
|
||||
---
|
||||
|
||||
## Visualization of Multipack with Flash Attention
|
||||
|
||||
@@ -1,4 +1,7 @@
|
||||
# NCCL
|
||||
---
|
||||
title: NCCL
|
||||
description: Troubleshooting NCCL issues
|
||||
---
|
||||
|
||||
NVIDIA NCCL is a library to facilitate and optimize multi-GPU communication operations, such as broadcast, all-gather, reduce, all-reduce, etc. Broadly, NCCL configuration is highly environment-specific and is configured via several [environment variables](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/env.html). A common NCCL-related problem occurs when a long-running operation times out causing the training process to abort:
|
||||
|
||||
@@ -1,29 +0,0 @@
|
||||
# Optimizers
|
||||
|
||||
Optimizers are an important component when training LLMs. Optimizers are responsible for updating the model's weights (parameters) based on the gradients computed during backpropagation.
|
||||
The goal of an optimizer is to minimize the loss function.
|
||||
|
||||
### Adam/AdamW Optimizers
|
||||
|
||||
```yaml
|
||||
adam_beta1: 0.9
|
||||
adam_beta2: 0.999
|
||||
adam_epsilon: 1e-8
|
||||
weight_decay: 0.0
|
||||
```
|
||||
|
||||
### GaLore Optimizer
|
||||
|
||||
https://huggingface.co/papers/2403.03507
|
||||
|
||||
```yaml
|
||||
optimizer: galore_adamw | galore_adamw_8bit | galore_adafactor
|
||||
optim_args:
|
||||
rank: 128
|
||||
update_proj_gap: 200
|
||||
scale: 0.25
|
||||
proj_type: std
|
||||
optim_target_modules:
|
||||
- mlp
|
||||
- attn
|
||||
```
|
||||
@@ -1,4 +1,7 @@
|
||||
# RLHF (Beta)
|
||||
---
|
||||
title: "RLHF (Beta)"
|
||||
description: "Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human feedback."
|
||||
---
|
||||
|
||||
### Overview
|
||||
|
||||
@@ -36,7 +36,7 @@ wandb_log_model:
|
||||
gradient_accumulation_steps: 4
|
||||
micro_batch_size: 4
|
||||
num_epochs: 4
|
||||
optimizer: paged_adamw_8bit
|
||||
optimizer: adamw_torch
|
||||
lr_scheduler: cosine
|
||||
learning_rate: 0.00001
|
||||
|
||||
@@ -66,5 +66,11 @@ weight_decay: 0.0
|
||||
fsdp:
|
||||
- full_shard
|
||||
fsdp_config:
|
||||
fsdp_limit_all_gathers: true
|
||||
fsdp_sync_module_states: true
|
||||
fsdp_offload_params: true
|
||||
fsdp_use_orig_params: false
|
||||
fsdp_cpu_ram_efficient_loading: true
|
||||
fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
|
||||
fsdp_state_dict_type: SHARDED_STATE_DICT
|
||||
special_tokens:
|
||||
|
||||
@@ -1,12 +0,0 @@
|
||||
# Description
|
||||
This repository presents an in-depth guide for fine-tuning Mistral-7b or any other compatible model using Axolotl, tailored specifically for chatbot development. It streamlines the process of fine-tuning and uploading the enhanced model to HuggingFace 🤗, thereby serving as an invaluable tool for developers in the AI and chatbot domain.
|
||||
|
||||
**What’s Inside:**
|
||||
|
||||
Beginner-Friendly Instructions: Comprehensive steps to guide you through fine-tuning your chosen model, including details on the data structure (jsonl), configuration, and the code itself.
|
||||
|
||||
Hardware Utilized: For reference, the fine-tuning in this guide was performed using 4x NVIDIA GeForce RTX 3090 (rented 2.1.2-cuda12.1-cudnn8-devel).
|
||||
|
||||
**Uploading to HuggingFace 🤗:**
|
||||
To upload your fine-tuned model to Hugging Face, include the following files:
|
||||
|
||||
@@ -1,970 +0,0 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"id": "3fe31229-8f6b-48bc-a86d-af8e5466d11c",
|
||||
"metadata": {
|
||||
"scrolled": true
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"GPU available? True\n",
|
||||
"BF16 is supported? True\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# Check if GPU is available I used 4x NVIDIA GeForce RTX 3090 (rented 2.1.2-cuda12.1-cudnn8-devel)\n",
|
||||
"import torch\n",
|
||||
"print('GPU available?', torch.cuda.is_available())\n",
|
||||
"print('BF16 is supported?', torch.cuda.is_bf16_supported())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"id": "1dee845b-f3cb-4b1e-bdd9-1a918eac140b",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Collecting huggingface_hub\n",
|
||||
" Downloading huggingface_hub-0.20.1-py3-none-any.whl.metadata (12 kB)\n",
|
||||
"Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (3.9.0)\n",
|
||||
"Requirement already satisfied: fsspec>=2023.5.0 in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (2023.10.0)\n",
|
||||
"Requirement already satisfied: requests in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (2.31.0)\n",
|
||||
"Requirement already satisfied: tqdm>=4.42.1 in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (4.65.0)\n",
|
||||
"Requirement already satisfied: pyyaml>=5.1 in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (6.0.1)\n",
|
||||
"Requirement already satisfied: typing-extensions>=3.7.4.3 in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (4.7.1)\n",
|
||||
"Requirement already satisfied: packaging>=20.9 in /opt/conda/lib/python3.10/site-packages (from huggingface_hub) (23.1)\n",
|
||||
"Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface_hub) (2.0.4)\n",
|
||||
"Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface_hub) (3.4)\n",
|
||||
"Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface_hub) (1.26.18)\n",
|
||||
"Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface_hub) (2023.7.22)\n",
|
||||
"Downloading huggingface_hub-0.20.1-py3-none-any.whl (330 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m330.1/330.1 kB\u001b[0m \u001b[31m8.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m\n",
|
||||
"\u001b[?25hInstalling collected packages: huggingface_hub\n",
|
||||
"Successfully installed huggingface_hub-0.20.1\n",
|
||||
"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
|
||||
"\u001b[0m"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"!pip install huggingface_hub"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 4,
|
||||
"id": "88731672-9050-4034-8266-11aaace2a44e",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from huggingface_hub import notebook_login"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 5,
|
||||
"id": "6b5aa7d7-3b18-4c14-afd4-043c2c545259",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"data": {
|
||||
"application/vnd.jupyter.widget-view+json": {
|
||||
"model_id": "60df98d7b0294289aad8b6c8cd023c3b",
|
||||
"version_major": 2,
|
||||
"version_minor": 0
|
||||
},
|
||||
"text/plain": [
|
||||
"VBox(children=(HTML(value='<center> <img\\nsrc=https://huggingface.co/front/assets/huggingface_logo-noborder.sv…"
|
||||
]
|
||||
},
|
||||
"metadata": {},
|
||||
"output_type": "display_data"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"#Login to huggingface so you can push the model to hub later\n",
|
||||
"import sys\n",
|
||||
"stdout = sys.stdout\n",
|
||||
"notebook_login()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"id": "b74d0635-5033-4494-b7bd-ff6822103d93",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#I noticed that when you use notebook_login() nothing gets printed after so we use sys \n",
|
||||
"sys.stdout = stdout"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 7,
|
||||
"id": "e3c3b088-45e7-484b-ae39-66beabc48da8",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Cloning into 'axolotl'...\n",
|
||||
"remote: Enumerating objects: 235, done.\u001b[K\n",
|
||||
"remote: Counting objects: 100% (235/235), done.\u001b[K\n",
|
||||
"remote: Compressing objects: 100% (207/207), done.\u001b[K\n",
|
||||
"remote: Total 235 (delta 48), reused 123 (delta 13), pack-reused 0\u001b[K\n",
|
||||
"Receiving objects: 100% (235/235), 1.46 MiB | 11.65 MiB/s, done.\n",
|
||||
"Resolving deltas: 100% (48/48), done.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"#axolotl\n",
|
||||
"!git clone -b main --depth 1 https://github.com/OpenAccess-AI-Collective/axolotl"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 8,
|
||||
"id": "66927751-4fd6-4477-97fc-6ab08c9d9a74",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"/axolotl\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"cd axolotl"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 9,
|
||||
"id": "fcccf8da-353b-4d70-8f55-5cfe08c7e6b9",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Requirement already satisfied: packaging in /opt/conda/lib/python3.10/site-packages (23.1)\n",
|
||||
"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
|
||||
"\u001b[0mObtaining file:///axolotl\n",
|
||||
" Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hCollecting auto-gptq==0.5.1\n",
|
||||
" Downloading auto_gptq-0.5.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (20 kB)\n",
|
||||
"Requirement already satisfied: packaging in /opt/conda/lib/python3.10/site-packages (23.1)\n",
|
||||
"Collecting peft==0.6.0\n",
|
||||
" Downloading peft-0.6.0-py3-none-any.whl.metadata (23 kB)\n",
|
||||
"Collecting transformers==4.36.2\n",
|
||||
" Downloading transformers-4.36.2-py3-none-any.whl.metadata (126 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m126.8/126.8 kB\u001b[0m \u001b[31m9.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting tokenizers==0.15.0\n",
|
||||
" Downloading tokenizers-0.15.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.7 kB)\n",
|
||||
"Collecting bitsandbytes>=0.41.1\n",
|
||||
" Downloading bitsandbytes-0.41.3.post2-py3-none-any.whl.metadata (9.8 kB)\n",
|
||||
"Collecting accelerate==0.24.1\n",
|
||||
" Downloading accelerate-0.24.1-py3-none-any.whl.metadata (18 kB)\n",
|
||||
"Collecting addict\n",
|
||||
" Downloading addict-2.4.0-py3-none-any.whl (3.8 kB)\n",
|
||||
"Collecting fire\n",
|
||||
" Downloading fire-0.5.0.tar.gz (88 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m88.3/88.3 kB\u001b[0m \u001b[31m28.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25h Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hRequirement already satisfied: PyYAML>=6.0 in /opt/conda/lib/python3.10/site-packages (6.0.1)\n",
|
||||
"Collecting datasets>=2.15.0\n",
|
||||
" Downloading datasets-2.16.0-py3-none-any.whl.metadata (20 kB)\n",
|
||||
"Collecting sentencepiece\n",
|
||||
" Downloading sentencepiece-0.1.99-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.3/1.3 MB\u001b[0m \u001b[31m47.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting wandb\n",
|
||||
" Downloading wandb-0.16.1-py3-none-any.whl.metadata (9.8 kB)\n",
|
||||
"Collecting einops\n",
|
||||
" Downloading einops-0.7.0-py3-none-any.whl.metadata (13 kB)\n",
|
||||
"Collecting optimum==1.13.2\n",
|
||||
" Downloading optimum-1.13.2.tar.gz (300 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m301.0/301.0 kB\u001b[0m \u001b[31m72.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25h Installing build dependencies ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Getting requirements to build wheel ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Preparing metadata (pyproject.toml) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hCollecting hf_transfer\n",
|
||||
" Downloading hf_transfer-0.1.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (1.5 kB)\n",
|
||||
"Collecting colorama\n",
|
||||
" Downloading colorama-0.4.6-py2.py3-none-any.whl (25 kB)\n",
|
||||
"Collecting numba\n",
|
||||
" Downloading numba-0.58.1-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (2.7 kB)\n",
|
||||
"Requirement already satisfied: numpy>=1.24.4 in /opt/conda/lib/python3.10/site-packages (1.26.0)\n",
|
||||
"Collecting bert-score==0.3.13\n",
|
||||
" Downloading bert_score-0.3.13-py3-none-any.whl (61 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m61.1/61.1 kB\u001b[0m \u001b[31m20.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting evaluate==0.4.0\n",
|
||||
" Downloading evaluate-0.4.0-py3-none-any.whl (81 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m81.4/81.4 kB\u001b[0m \u001b[31m26.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting rouge-score==0.1.2\n",
|
||||
" Downloading rouge_score-0.1.2.tar.gz (17 kB)\n",
|
||||
" Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hCollecting scipy\n",
|
||||
" Downloading scipy-1.11.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (60 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m60.4/60.4 kB\u001b[0m \u001b[31m17.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting scikit-learn==1.2.2\n",
|
||||
" Downloading scikit_learn-1.2.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (9.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m9.6/9.6 MB\u001b[0m \u001b[31m83.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m00:01\u001b[0m00:01\u001b[0mm\n",
|
||||
"\u001b[?25hCollecting pynvml\n",
|
||||
" Downloading pynvml-11.5.0-py3-none-any.whl (53 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m53.1/53.1 kB\u001b[0m \u001b[31m13.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting art\n",
|
||||
" Downloading art-6.1-py3-none-any.whl.metadata (69 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m69.9/69.9 kB\u001b[0m \u001b[31m21.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting fschat==0.2.34\n",
|
||||
" Downloading fschat-0.2.34-py3-none-any.whl.metadata (20 kB)\n",
|
||||
"Collecting gradio==3.50.2\n",
|
||||
" Downloading gradio-3.50.2-py3-none-any.whl.metadata (17 kB)\n",
|
||||
"Collecting tensorboard\n",
|
||||
" Downloading tensorboard-2.15.1-py3-none-any.whl.metadata (1.7 kB)\n",
|
||||
"Collecting s3fs\n",
|
||||
" Downloading s3fs-2023.12.2-py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
"Collecting gcsfs\n",
|
||||
" Downloading gcsfs-2023.12.2.post1-py2.py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
"Collecting xformers==0.0.23\n",
|
||||
" Downloading xformers-0.0.23-cp310-cp310-manylinux2014_x86_64.whl.metadata (1.0 kB)\n",
|
||||
"Collecting deepspeed\n",
|
||||
" Downloading deepspeed-0.12.6.tar.gz (1.2 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.2/1.2 MB\u001b[0m \u001b[31m109.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25h Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hCollecting flash-attn==2.3.3\n",
|
||||
" Downloading flash_attn-2.3.3.tar.gz (2.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.3/2.3 MB\u001b[0m \u001b[31m111.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25h Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hRequirement already satisfied: psutil in /opt/conda/lib/python3.10/site-packages (from accelerate==0.24.1) (5.9.0)\n",
|
||||
"Requirement already satisfied: torch>=1.10.0 in /opt/conda/lib/python3.10/site-packages (from accelerate==0.24.1) (2.1.1)\n",
|
||||
"Requirement already satisfied: huggingface-hub in /opt/conda/lib/python3.10/site-packages (from accelerate==0.24.1) (0.20.1)\n",
|
||||
"Collecting rouge (from auto-gptq==0.5.1)\n",
|
||||
" Downloading rouge-1.0.1-py3-none-any.whl (13 kB)\n",
|
||||
"Collecting gekko (from auto-gptq==0.5.1)\n",
|
||||
" Downloading gekko-1.0.6-py3-none-any.whl (12.2 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m12.2/12.2 MB\u001b[0m \u001b[31m77.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m00:01\u001b[0m0:01\u001b[0m\n",
|
||||
"\u001b[?25hCollecting safetensors (from auto-gptq==0.5.1)\n",
|
||||
" Downloading safetensors-0.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (3.8 kB)\n",
|
||||
"Requirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from auto-gptq==0.5.1) (4.65.0)\n",
|
||||
"Collecting pandas>=1.0.1 (from bert-score==0.3.13)\n",
|
||||
" Downloading pandas-2.1.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (18 kB)\n",
|
||||
"Requirement already satisfied: requests in /opt/conda/lib/python3.10/site-packages (from bert-score==0.3.13) (2.31.0)\n",
|
||||
"Collecting matplotlib (from bert-score==0.3.13)\n",
|
||||
" Downloading matplotlib-3.8.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (5.8 kB)\n",
|
||||
"Collecting dill (from evaluate==0.4.0)\n",
|
||||
" Downloading dill-0.3.7-py3-none-any.whl.metadata (9.9 kB)\n",
|
||||
"Collecting xxhash (from evaluate==0.4.0)\n",
|
||||
" Downloading xxhash-3.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (12 kB)\n",
|
||||
"Collecting multiprocess (from evaluate==0.4.0)\n",
|
||||
" Downloading multiprocess-0.70.15-py310-none-any.whl.metadata (7.2 kB)\n",
|
||||
"Requirement already satisfied: fsspec>=2021.05.0 in /opt/conda/lib/python3.10/site-packages (from fsspec[http]>=2021.05.0->evaluate==0.4.0) (2023.10.0)\n",
|
||||
"Collecting responses<0.19 (from evaluate==0.4.0)\n",
|
||||
" Downloading responses-0.18.0-py3-none-any.whl (38 kB)\n",
|
||||
"Collecting ninja (from flash-attn==2.3.3)\n",
|
||||
" Downloading ninja-1.11.1.1-py2.py3-none-manylinux1_x86_64.manylinux_2_5_x86_64.whl.metadata (5.3 kB)\n",
|
||||
"Collecting aiohttp (from fschat==0.2.34)\n",
|
||||
" Downloading aiohttp-3.9.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.4 kB)\n",
|
||||
"Collecting fastapi (from fschat==0.2.34)\n",
|
||||
" Downloading fastapi-0.108.0-py3-none-any.whl.metadata (24 kB)\n",
|
||||
"Collecting httpx (from fschat==0.2.34)\n",
|
||||
" Downloading httpx-0.26.0-py3-none-any.whl.metadata (7.6 kB)\n",
|
||||
"Collecting markdown2[all] (from fschat==0.2.34)\n",
|
||||
" Downloading markdown2-2.4.12-py2.py3-none-any.whl.metadata (2.0 kB)\n",
|
||||
"Collecting nh3 (from fschat==0.2.34)\n",
|
||||
" Downloading nh3-0.2.15-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (1.7 kB)\n",
|
||||
"Requirement already satisfied: prompt-toolkit>=3.0.0 in /opt/conda/lib/python3.10/site-packages (from fschat==0.2.34) (3.0.36)\n",
|
||||
"Collecting pydantic<2,>=1 (from fschat==0.2.34)\n",
|
||||
" Downloading pydantic-1.10.13-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (149 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m149.6/149.6 kB\u001b[0m \u001b[31m42.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting rich>=10.0.0 (from fschat==0.2.34)\n",
|
||||
" Downloading rich-13.7.0-py3-none-any.whl.metadata (18 kB)\n",
|
||||
"Collecting shortuuid (from fschat==0.2.34)\n",
|
||||
" Downloading shortuuid-1.0.11-py3-none-any.whl (10 kB)\n",
|
||||
"Collecting tiktoken (from fschat==0.2.34)\n",
|
||||
" Downloading tiktoken-0.5.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.6 kB)\n",
|
||||
"Collecting uvicorn (from fschat==0.2.34)\n",
|
||||
" Downloading uvicorn-0.25.0-py3-none-any.whl.metadata (6.4 kB)\n",
|
||||
"Collecting aiofiles<24.0,>=22.0 (from gradio==3.50.2)\n",
|
||||
" Downloading aiofiles-23.2.1-py3-none-any.whl.metadata (9.7 kB)\n",
|
||||
"Collecting altair<6.0,>=4.2.0 (from gradio==3.50.2)\n",
|
||||
" Downloading altair-5.2.0-py3-none-any.whl.metadata (8.7 kB)\n",
|
||||
"Collecting ffmpy (from gradio==3.50.2)\n",
|
||||
" Downloading ffmpy-0.3.1.tar.gz (5.5 kB)\n",
|
||||
" Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hCollecting gradio-client==0.6.1 (from gradio==3.50.2)\n",
|
||||
" Downloading gradio_client-0.6.1-py3-none-any.whl.metadata (7.1 kB)\n",
|
||||
"Collecting importlib-resources<7.0,>=1.3 (from gradio==3.50.2)\n",
|
||||
" Downloading importlib_resources-6.1.1-py3-none-any.whl.metadata (4.1 kB)\n",
|
||||
"Requirement already satisfied: jinja2<4.0 in /opt/conda/lib/python3.10/site-packages (from gradio==3.50.2) (3.1.2)\n",
|
||||
"Requirement already satisfied: markupsafe~=2.0 in /opt/conda/lib/python3.10/site-packages (from gradio==3.50.2) (2.1.1)\n",
|
||||
"Collecting orjson~=3.0 (from gradio==3.50.2)\n",
|
||||
" Downloading orjson-3.9.10-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (49 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m49.3/49.3 kB\u001b[0m \u001b[31m14.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: pillow<11.0,>=8.0 in /opt/conda/lib/python3.10/site-packages (from gradio==3.50.2) (10.0.1)\n",
|
||||
"Collecting pydub (from gradio==3.50.2)\n",
|
||||
" Downloading pydub-0.25.1-py2.py3-none-any.whl (32 kB)\n",
|
||||
"Collecting python-multipart (from gradio==3.50.2)\n",
|
||||
" Downloading python_multipart-0.0.6-py3-none-any.whl (45 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m45.7/45.7 kB\u001b[0m \u001b[31m13.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting semantic-version~=2.0 (from gradio==3.50.2)\n",
|
||||
" Downloading semantic_version-2.10.0-py2.py3-none-any.whl (15 kB)\n",
|
||||
"Requirement already satisfied: typing-extensions~=4.0 in /opt/conda/lib/python3.10/site-packages (from gradio==3.50.2) (4.7.1)\n",
|
||||
"Collecting websockets<12.0,>=10.0 (from gradio==3.50.2)\n",
|
||||
" Downloading websockets-11.0.3-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (129 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m129.9/129.9 kB\u001b[0m \u001b[31m30.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting coloredlogs (from optimum==1.13.2)\n",
|
||||
" Downloading coloredlogs-15.0.1-py2.py3-none-any.whl (46 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m46.0/46.0 kB\u001b[0m \u001b[31m11.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: sympy in /opt/conda/lib/python3.10/site-packages (from optimum==1.13.2) (1.11.1)\n",
|
||||
"Collecting absl-py (from rouge-score==0.1.2)\n",
|
||||
" Downloading absl_py-2.0.0-py3-none-any.whl.metadata (2.3 kB)\n",
|
||||
"Collecting nltk (from rouge-score==0.1.2)\n",
|
||||
" Downloading nltk-3.8.1-py3-none-any.whl (1.5 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.5/1.5 MB\u001b[0m \u001b[31m90.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: six>=1.14.0 in /opt/conda/lib/python3.10/site-packages (from rouge-score==0.1.2) (1.16.0)\n",
|
||||
"Collecting joblib>=1.1.1 (from scikit-learn==1.2.2)\n",
|
||||
" Downloading joblib-1.3.2-py3-none-any.whl.metadata (5.4 kB)\n",
|
||||
"Collecting threadpoolctl>=2.0.0 (from scikit-learn==1.2.2)\n",
|
||||
" Downloading threadpoolctl-3.2.0-py3-none-any.whl.metadata (10.0 kB)\n",
|
||||
"Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from transformers==4.36.2) (3.9.0)\n",
|
||||
"Collecting regex!=2019.12.17 (from transformers==4.36.2)\n",
|
||||
" Downloading regex-2023.12.25-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (40 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m40.9/40.9 kB\u001b[0m \u001b[31m12.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: networkx in /opt/conda/lib/python3.10/site-packages (from torch>=1.10.0->accelerate==0.24.1) (3.1)\n",
|
||||
"Collecting pyarrow>=8.0.0 (from datasets>=2.15.0)\n",
|
||||
" Downloading pyarrow-14.0.2-cp310-cp310-manylinux_2_28_x86_64.whl.metadata (3.0 kB)\n",
|
||||
"Collecting pyarrow-hotfix (from datasets>=2.15.0)\n",
|
||||
" Downloading pyarrow_hotfix-0.6-py3-none-any.whl.metadata (3.6 kB)\n",
|
||||
"Collecting hjson (from deepspeed)\n",
|
||||
" Downloading hjson-3.1.0-py3-none-any.whl (54 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m54.0/54.0 kB\u001b[0m \u001b[31m19.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting py-cpuinfo (from deepspeed)\n",
|
||||
" Downloading py_cpuinfo-9.0.0-py3-none-any.whl (22 kB)\n",
|
||||
"Collecting termcolor (from fire)\n",
|
||||
" Downloading termcolor-2.4.0-py3-none-any.whl.metadata (6.1 kB)\n",
|
||||
"Requirement already satisfied: decorator>4.1.2 in /opt/conda/lib/python3.10/site-packages (from gcsfs) (5.1.1)\n",
|
||||
"INFO: pip is looking at multiple versions of gcsfs to determine which version is compatible with other requirements. This could take a while.\n",
|
||||
"Collecting gcsfs\n",
|
||||
" Downloading gcsfs-2023.12.1-py2.py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
" Downloading gcsfs-2023.12.0-py2.py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
" Downloading gcsfs-2023.10.0-py2.py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
"Collecting google-auth>=1.2 (from gcsfs)\n",
|
||||
" Downloading google_auth-2.25.2-py2.py3-none-any.whl.metadata (4.7 kB)\n",
|
||||
"Collecting google-auth-oauthlib (from gcsfs)\n",
|
||||
" Downloading google_auth_oauthlib-1.2.0-py2.py3-none-any.whl.metadata (2.7 kB)\n",
|
||||
"Collecting google-cloud-storage (from gcsfs)\n",
|
||||
" Downloading google_cloud_storage-2.14.0-py2.py3-none-any.whl.metadata (6.1 kB)\n",
|
||||
"Collecting llvmlite<0.42,>=0.41.0dev0 (from numba)\n",
|
||||
" Downloading llvmlite-0.41.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.8 kB)\n",
|
||||
"Collecting aiobotocore<3.0.0,>=2.5.4 (from s3fs)\n",
|
||||
" Downloading aiobotocore-2.9.0-py3-none-any.whl.metadata (20 kB)\n",
|
||||
"INFO: pip is looking at multiple versions of s3fs to determine which version is compatible with other requirements. This could take a while.\n",
|
||||
"Collecting s3fs\n",
|
||||
" Downloading s3fs-2023.12.1-py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
" Downloading s3fs-2023.10.0-py3-none-any.whl.metadata (1.6 kB)\n",
|
||||
"Collecting aiobotocore~=2.7.0 (from s3fs)\n",
|
||||
" Downloading aiobotocore-2.7.0-py3-none-any.whl.metadata (20 kB)\n",
|
||||
"Collecting grpcio>=1.48.2 (from tensorboard)\n",
|
||||
" Downloading grpcio-1.60.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.0 kB)\n",
|
||||
"Collecting markdown>=2.6.8 (from tensorboard)\n",
|
||||
" Downloading Markdown-3.5.1-py3-none-any.whl.metadata (7.1 kB)\n",
|
||||
"Collecting protobuf<4.24,>=3.19.6 (from tensorboard)\n",
|
||||
" Downloading protobuf-4.23.4-cp37-abi3-manylinux2014_x86_64.whl.metadata (540 bytes)\n",
|
||||
"Requirement already satisfied: setuptools>=41.0.0 in /opt/conda/lib/python3.10/site-packages (from tensorboard) (68.0.0)\n",
|
||||
"Collecting tensorboard-data-server<0.8.0,>=0.7.0 (from tensorboard)\n",
|
||||
" Downloading tensorboard_data_server-0.7.2-py3-none-manylinux_2_31_x86_64.whl.metadata (1.1 kB)\n",
|
||||
"Collecting werkzeug>=1.0.1 (from tensorboard)\n",
|
||||
" Downloading werkzeug-3.0.1-py3-none-any.whl.metadata (4.1 kB)\n",
|
||||
"Requirement already satisfied: Click!=8.0.0,>=7.1 in /opt/conda/lib/python3.10/site-packages (from wandb) (8.1.7)\n",
|
||||
"Collecting GitPython!=3.1.29,>=1.0.0 (from wandb)\n",
|
||||
" Downloading GitPython-3.1.40-py3-none-any.whl.metadata (12 kB)\n",
|
||||
"Collecting sentry-sdk>=1.0.0 (from wandb)\n",
|
||||
" Downloading sentry_sdk-1.39.1-py2.py3-none-any.whl.metadata (9.7 kB)\n",
|
||||
"Collecting docker-pycreds>=0.4.0 (from wandb)\n",
|
||||
" Downloading docker_pycreds-0.4.0-py2.py3-none-any.whl (9.0 kB)\n",
|
||||
"Collecting setproctitle (from wandb)\n",
|
||||
" Downloading setproctitle-1.3.3-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (9.9 kB)\n",
|
||||
"Collecting appdirs>=1.4.3 (from wandb)\n",
|
||||
" Downloading appdirs-1.4.4-py2.py3-none-any.whl (9.6 kB)\n",
|
||||
"Collecting botocore<1.31.65,>=1.31.16 (from aiobotocore~=2.7.0->s3fs)\n",
|
||||
" Downloading botocore-1.31.64-py3-none-any.whl.metadata (6.1 kB)\n",
|
||||
"Collecting wrapt<2.0.0,>=1.10.10 (from aiobotocore~=2.7.0->s3fs)\n",
|
||||
" Downloading wrapt-1.16.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.6 kB)\n",
|
||||
"Collecting aioitertools<1.0.0,>=0.5.1 (from aiobotocore~=2.7.0->s3fs)\n",
|
||||
" Downloading aioitertools-0.11.0-py3-none-any.whl (23 kB)\n",
|
||||
"Requirement already satisfied: attrs>=17.3.0 in /opt/conda/lib/python3.10/site-packages (from aiohttp->fschat==0.2.34) (23.1.0)\n",
|
||||
"Collecting multidict<7.0,>=4.5 (from aiohttp->fschat==0.2.34)\n",
|
||||
" Downloading multidict-6.0.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (114 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m114.5/114.5 kB\u001b[0m \u001b[31m37.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting yarl<2.0,>=1.0 (from aiohttp->fschat==0.2.34)\n",
|
||||
" Downloading yarl-1.9.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (31 kB)\n",
|
||||
"Collecting frozenlist>=1.1.1 (from aiohttp->fschat==0.2.34)\n",
|
||||
" Downloading frozenlist-1.4.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (12 kB)\n",
|
||||
"Collecting aiosignal>=1.1.2 (from aiohttp->fschat==0.2.34)\n",
|
||||
" Downloading aiosignal-1.3.1-py3-none-any.whl (7.6 kB)\n",
|
||||
"Collecting async-timeout<5.0,>=4.0 (from aiohttp->fschat==0.2.34)\n",
|
||||
" Downloading async_timeout-4.0.3-py3-none-any.whl.metadata (4.2 kB)\n",
|
||||
"Requirement already satisfied: jsonschema>=3.0 in /opt/conda/lib/python3.10/site-packages (from altair<6.0,>=4.2.0->gradio==3.50.2) (4.20.0)\n",
|
||||
"Requirement already satisfied: toolz in /opt/conda/lib/python3.10/site-packages (from altair<6.0,>=4.2.0->gradio==3.50.2) (0.12.0)\n",
|
||||
"Collecting gitdb<5,>=4.0.1 (from GitPython!=3.1.29,>=1.0.0->wandb)\n",
|
||||
" Downloading gitdb-4.0.11-py3-none-any.whl.metadata (1.2 kB)\n",
|
||||
"Collecting cachetools<6.0,>=2.0.0 (from google-auth>=1.2->gcsfs)\n",
|
||||
" Downloading cachetools-5.3.2-py3-none-any.whl.metadata (5.2 kB)\n",
|
||||
"Collecting pyasn1-modules>=0.2.1 (from google-auth>=1.2->gcsfs)\n",
|
||||
" Downloading pyasn1_modules-0.3.0-py2.py3-none-any.whl (181 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m181.3/181.3 kB\u001b[0m \u001b[31m59.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting rsa<5,>=3.1.4 (from google-auth>=1.2->gcsfs)\n",
|
||||
" Downloading rsa-4.9-py3-none-any.whl (34 kB)\n",
|
||||
"Collecting requests-oauthlib>=0.7.0 (from google-auth-oauthlib->gcsfs)\n",
|
||||
" Downloading requests_oauthlib-1.3.1-py2.py3-none-any.whl (23 kB)\n",
|
||||
"Collecting contourpy>=1.0.1 (from matplotlib->bert-score==0.3.13)\n",
|
||||
" Downloading contourpy-1.2.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (5.8 kB)\n",
|
||||
"Collecting cycler>=0.10 (from matplotlib->bert-score==0.3.13)\n",
|
||||
" Downloading cycler-0.12.1-py3-none-any.whl.metadata (3.8 kB)\n",
|
||||
"Collecting fonttools>=4.22.0 (from matplotlib->bert-score==0.3.13)\n",
|
||||
" Downloading fonttools-4.47.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (157 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m157.2/157.2 kB\u001b[0m \u001b[31m41.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting kiwisolver>=1.3.1 (from matplotlib->bert-score==0.3.13)\n",
|
||||
" Downloading kiwisolver-1.4.5-cp310-cp310-manylinux_2_12_x86_64.manylinux2010_x86_64.whl.metadata (6.4 kB)\n",
|
||||
"Collecting pyparsing>=2.3.1 (from matplotlib->bert-score==0.3.13)\n",
|
||||
" Downloading pyparsing-3.1.1-py3-none-any.whl.metadata (5.1 kB)\n",
|
||||
"Requirement already satisfied: python-dateutil>=2.7 in /opt/conda/lib/python3.10/site-packages (from matplotlib->bert-score==0.3.13) (2.8.2)\n",
|
||||
"Requirement already satisfied: pytz>=2020.1 in /opt/conda/lib/python3.10/site-packages (from pandas>=1.0.1->bert-score==0.3.13) (2023.3.post1)\n",
|
||||
"Collecting tzdata>=2022.1 (from pandas>=1.0.1->bert-score==0.3.13)\n",
|
||||
" Downloading tzdata-2023.3-py2.py3-none-any.whl (341 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m341.8/341.8 kB\u001b[0m \u001b[31m72.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: wcwidth in /opt/conda/lib/python3.10/site-packages (from prompt-toolkit>=3.0.0->fschat==0.2.34) (0.2.5)\n",
|
||||
"Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests->bert-score==0.3.13) (2.0.4)\n",
|
||||
"Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests->bert-score==0.3.13) (3.4)\n",
|
||||
"Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests->bert-score==0.3.13) (1.26.18)\n",
|
||||
"Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests->bert-score==0.3.13) (2023.7.22)\n",
|
||||
"Collecting markdown-it-py>=2.2.0 (from rich>=10.0.0->fschat==0.2.34)\n",
|
||||
" Downloading markdown_it_py-3.0.0-py3-none-any.whl.metadata (6.9 kB)\n",
|
||||
"Requirement already satisfied: pygments<3.0.0,>=2.13.0 in /opt/conda/lib/python3.10/site-packages (from rich>=10.0.0->fschat==0.2.34) (2.15.1)\n",
|
||||
"Collecting h11>=0.8 (from uvicorn->fschat==0.2.34)\n",
|
||||
" Downloading h11-0.14.0-py3-none-any.whl (58 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m58.3/58.3 kB\u001b[0m \u001b[31m21.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting humanfriendly>=9.1 (from coloredlogs->optimum==1.13.2)\n",
|
||||
" Downloading humanfriendly-10.0-py2.py3-none-any.whl (86 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m86.8/86.8 kB\u001b[0m \u001b[31m27.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hCollecting starlette<0.33.0,>=0.29.0 (from fastapi->fschat==0.2.34)\n",
|
||||
" Downloading starlette-0.32.0.post1-py3-none-any.whl.metadata (5.8 kB)\n",
|
||||
"Collecting typing-extensions~=4.0 (from gradio==3.50.2)\n",
|
||||
" Downloading typing_extensions-4.9.0-py3-none-any.whl.metadata (3.0 kB)\n",
|
||||
"Collecting google-api-core!=2.0.*,!=2.1.*,!=2.2.*,!=2.3.0,<3.0.0dev,>=1.31.5 (from google-cloud-storage->gcsfs)\n",
|
||||
" Downloading google_api_core-2.15.0-py3-none-any.whl.metadata (2.7 kB)\n",
|
||||
"Collecting google-cloud-core<3.0dev,>=2.3.0 (from google-cloud-storage->gcsfs)\n",
|
||||
" Downloading google_cloud_core-2.4.1-py2.py3-none-any.whl.metadata (2.7 kB)\n",
|
||||
"Collecting google-resumable-media>=2.6.0 (from google-cloud-storage->gcsfs)\n",
|
||||
" Downloading google_resumable_media-2.7.0-py2.py3-none-any.whl.metadata (2.2 kB)\n",
|
||||
"Collecting google-crc32c<2.0dev,>=1.0 (from google-cloud-storage->gcsfs)\n",
|
||||
" Downloading google_crc32c-1.5.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (32 kB)\n",
|
||||
"Requirement already satisfied: anyio in /opt/conda/lib/python3.10/site-packages (from httpx->fschat==0.2.34) (4.2.0)\n",
|
||||
"Collecting httpcore==1.* (from httpx->fschat==0.2.34)\n",
|
||||
" Downloading httpcore-1.0.2-py3-none-any.whl.metadata (20 kB)\n",
|
||||
"Requirement already satisfied: sniffio in /opt/conda/lib/python3.10/site-packages (from httpx->fschat==0.2.34) (1.3.0)\n",
|
||||
"Collecting wavedrom (from markdown2[all]->fschat==0.2.34)\n",
|
||||
" Downloading wavedrom-2.0.3.post3.tar.gz (137 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m137.7/137.7 kB\u001b[0m \u001b[31m47.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25h Preparing metadata (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hRequirement already satisfied: mpmath>=0.19 in /opt/conda/lib/python3.10/site-packages (from sympy->optimum==1.13.2) (1.3.0)\n",
|
||||
"Collecting jmespath<2.0.0,>=0.7.1 (from botocore<1.31.65,>=1.31.16->aiobotocore~=2.7.0->s3fs)\n",
|
||||
" Downloading jmespath-1.0.1-py3-none-any.whl (20 kB)\n",
|
||||
"Collecting smmap<6,>=3.0.1 (from gitdb<5,>=4.0.1->GitPython!=3.1.29,>=1.0.0->wandb)\n",
|
||||
" Downloading smmap-5.0.1-py3-none-any.whl.metadata (4.3 kB)\n",
|
||||
"Collecting googleapis-common-protos<2.0.dev0,>=1.56.2 (from google-api-core!=2.0.*,!=2.1.*,!=2.2.*,!=2.3.0,<3.0.0dev,>=1.31.5->google-cloud-storage->gcsfs)\n",
|
||||
" Downloading googleapis_common_protos-1.62.0-py2.py3-none-any.whl.metadata (1.5 kB)\n",
|
||||
"Requirement already satisfied: jsonschema-specifications>=2023.03.6 in /opt/conda/lib/python3.10/site-packages (from jsonschema>=3.0->altair<6.0,>=4.2.0->gradio==3.50.2) (2023.12.1)\n",
|
||||
"Requirement already satisfied: referencing>=0.28.4 in /opt/conda/lib/python3.10/site-packages (from jsonschema>=3.0->altair<6.0,>=4.2.0->gradio==3.50.2) (0.32.0)\n",
|
||||
"Requirement already satisfied: rpds-py>=0.7.1 in /opt/conda/lib/python3.10/site-packages (from jsonschema>=3.0->altair<6.0,>=4.2.0->gradio==3.50.2) (0.15.2)\n",
|
||||
"Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=10.0.0->fschat==0.2.34)\n",
|
||||
" Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB)\n",
|
||||
"Collecting pyasn1<0.6.0,>=0.4.6 (from pyasn1-modules>=0.2.1->google-auth>=1.2->gcsfs)\n",
|
||||
" Downloading pyasn1-0.5.1-py2.py3-none-any.whl.metadata (8.6 kB)\n",
|
||||
"Collecting oauthlib>=3.0.0 (from requests-oauthlib>=0.7.0->google-auth-oauthlib->gcsfs)\n",
|
||||
" Downloading oauthlib-3.2.2-py3-none-any.whl (151 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m151.7/151.7 kB\u001b[0m \u001b[31m50.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hRequirement already satisfied: exceptiongroup>=1.0.2 in /opt/conda/lib/python3.10/site-packages (from anyio->httpx->fschat==0.2.34) (1.0.4)\n",
|
||||
"Collecting svgwrite (from wavedrom->markdown2[all]->fschat==0.2.34)\n",
|
||||
" Downloading svgwrite-1.4.3-py3-none-any.whl (67 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m67.1/67.1 kB\u001b[0m \u001b[31m21.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading accelerate-0.24.1-py3-none-any.whl (261 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m261.4/261.4 kB\u001b[0m \u001b[31m53.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading auto_gptq-0.5.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (4.8 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m4.8/4.8 MB\u001b[0m \u001b[31m89.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mta \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading fschat-0.2.34-py3-none-any.whl (220 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m220.1/220.1 kB\u001b[0m \u001b[31m63.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading gradio-3.50.2-py3-none-any.whl (20.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m20.3/20.3 MB\u001b[0m \u001b[31m82.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading peft-0.6.0-py3-none-any.whl (134 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m134.9/134.9 kB\u001b[0m \u001b[31m40.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading tokenizers-0.15.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.8 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.8/3.8 MB\u001b[0m \u001b[31m87.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mta \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading transformers-4.36.2-py3-none-any.whl (8.2 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m8.2/8.2 MB\u001b[0m \u001b[31m90.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mta \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading xformers-0.0.23-cp310-cp310-manylinux2014_x86_64.whl (213.0 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m213.0/213.0 MB\u001b[0m \u001b[31m36.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading gradio_client-0.6.1-py3-none-any.whl (299 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m299.2/299.2 kB\u001b[0m \u001b[31m64.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading bitsandbytes-0.41.3.post2-py3-none-any.whl (92.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m92.6/92.6 MB\u001b[0m \u001b[31m56.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading datasets-2.16.0-py3-none-any.whl (507 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m507.1/507.1 kB\u001b[0m \u001b[31m87.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading scipy-1.11.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (36.4 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m36.4/36.4 MB\u001b[0m \u001b[31m77.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading art-6.1-py3-none-any.whl (599 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m599.8/599.8 kB\u001b[0m \u001b[31m96.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading einops-0.7.0-py3-none-any.whl (44 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m44.6/44.6 kB\u001b[0m \u001b[31m13.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading gcsfs-2023.10.0-py2.py3-none-any.whl (33 kB)\n",
|
||||
"Downloading hf_transfer-0.1.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.9 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.9/3.9 MB\u001b[0m \u001b[31m99.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading numba-0.58.1-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (3.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.6/3.6 MB\u001b[0m \u001b[31m100.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading s3fs-2023.10.0-py3-none-any.whl (28 kB)\n",
|
||||
"Downloading tensorboard-2.15.1-py3-none-any.whl (5.5 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m5.5/5.5 MB\u001b[0m \u001b[31m96.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0mta \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading wandb-0.16.1-py3-none-any.whl (2.1 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.1/2.1 MB\u001b[0m \u001b[31m99.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading absl_py-2.0.0-py3-none-any.whl (130 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m130.2/130.2 kB\u001b[0m \u001b[31m36.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading aiobotocore-2.7.0-py3-none-any.whl (73 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m73.5/73.5 kB\u001b[0m \u001b[31m25.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading aiofiles-23.2.1-py3-none-any.whl (15 kB)\n",
|
||||
"Downloading aiohttp-3.9.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.2 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.2/1.2 MB\u001b[0m \u001b[31m99.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading altair-5.2.0-py3-none-any.whl (996 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m996.9/996.9 kB\u001b[0m \u001b[31m110.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading dill-0.3.7-py3-none-any.whl (115 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m115.3/115.3 kB\u001b[0m \u001b[31m34.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading GitPython-3.1.40-py3-none-any.whl (190 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m190.6/190.6 kB\u001b[0m \u001b[31m47.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading google_auth-2.25.2-py2.py3-none-any.whl (184 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m184.2/184.2 kB\u001b[0m \u001b[31m44.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading google_auth_oauthlib-1.2.0-py2.py3-none-any.whl (24 kB)\n",
|
||||
"Downloading grpcio-1.60.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (5.4 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m5.4/5.4 MB\u001b[0m \u001b[31m102.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0ma \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading importlib_resources-6.1.1-py3-none-any.whl (33 kB)\n",
|
||||
"Downloading joblib-1.3.2-py3-none-any.whl (302 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m302.2/302.2 kB\u001b[0m \u001b[31m64.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading llvmlite-0.41.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (43.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m43.6/43.6 MB\u001b[0m \u001b[31m74.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading Markdown-3.5.1-py3-none-any.whl (102 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m102.2/102.2 kB\u001b[0m \u001b[31m34.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading matplotlib-3.8.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (11.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m11.6/11.6 MB\u001b[0m \u001b[31m99.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m0:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading orjson-3.9.10-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (138 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m138.7/138.7 kB\u001b[0m \u001b[31m38.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pandas-2.1.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (12.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m12.3/12.3 MB\u001b[0m \u001b[31m96.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m0:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading protobuf-4.23.4-cp37-abi3-manylinux2014_x86_64.whl (304 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m304.5/304.5 kB\u001b[0m \u001b[31m68.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pyarrow-14.0.2-cp310-cp310-manylinux_2_28_x86_64.whl (38.0 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m38.0/38.0 MB\u001b[0m \u001b[31m78.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pydantic-1.10.13-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.1 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.1/3.1 MB\u001b[0m \u001b[31m95.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading regex-2023.12.25-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (773 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m774.0/774.0 kB\u001b[0m \u001b[31m116.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading rich-13.7.0-py3-none-any.whl (240 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m240.6/240.6 kB\u001b[0m \u001b[31m59.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading safetensors-0.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.3/1.3 MB\u001b[0m \u001b[31m102.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading sentry_sdk-1.39.1-py2.py3-none-any.whl (254 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m254.1/254.1 kB\u001b[0m \u001b[31m71.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading tensorboard_data_server-0.7.2-py3-none-manylinux_2_31_x86_64.whl (6.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m6.6/6.6 MB\u001b[0m \u001b[31m104.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0ma \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading threadpoolctl-3.2.0-py3-none-any.whl (15 kB)\n",
|
||||
"Downloading uvicorn-0.25.0-py3-none-any.whl (60 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m60.3/60.3 kB\u001b[0m \u001b[31m19.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading werkzeug-3.0.1-py3-none-any.whl (226 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m226.7/226.7 kB\u001b[0m \u001b[31m67.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading fastapi-0.108.0-py3-none-any.whl (92 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m92.0/92.0 kB\u001b[0m \u001b[31m33.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading typing_extensions-4.9.0-py3-none-any.whl (32 kB)\n",
|
||||
"Downloading google_cloud_storage-2.14.0-py2.py3-none-any.whl (121 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m121.6/121.6 kB\u001b[0m \u001b[31m36.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading httpx-0.26.0-py3-none-any.whl (75 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m75.9/75.9 kB\u001b[0m \u001b[31m24.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading httpcore-1.0.2-py3-none-any.whl (76 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m76.9/76.9 kB\u001b[0m \u001b[31m28.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading multiprocess-0.70.15-py310-none-any.whl (134 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m134.8/134.8 kB\u001b[0m \u001b[31m48.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading nh3-0.2.15-cp37-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (1.7 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.7/1.7 MB\u001b[0m \u001b[31m108.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading ninja-1.11.1.1-py2.py3-none-manylinux1_x86_64.manylinux_2_5_x86_64.whl (307 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m307.2/307.2 kB\u001b[0m \u001b[31m66.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pyarrow_hotfix-0.6-py3-none-any.whl (7.9 kB)\n",
|
||||
"Downloading setproctitle-1.3.3-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (30 kB)\n",
|
||||
"Downloading termcolor-2.4.0-py3-none-any.whl (7.7 kB)\n",
|
||||
"Downloading tiktoken-0.5.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (2.0 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.0/2.0 MB\u001b[0m \u001b[31m101.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading xxhash-3.4.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (194 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m194.1/194.1 kB\u001b[0m \u001b[31m44.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading async_timeout-4.0.3-py3-none-any.whl (5.7 kB)\n",
|
||||
"Downloading botocore-1.31.64-py3-none-any.whl (11.3 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m11.3/11.3 MB\u001b[0m \u001b[31m98.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m:00:01\u001b[0m0:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading cachetools-5.3.2-py3-none-any.whl (9.3 kB)\n",
|
||||
"Downloading contourpy-1.2.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (310 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m310.7/310.7 kB\u001b[0m \u001b[31m69.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading cycler-0.12.1-py3-none-any.whl (8.3 kB)\n",
|
||||
"Downloading fonttools-4.47.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (4.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m4.6/4.6 MB\u001b[0m \u001b[31m102.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0ma \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hDownloading frozenlist-1.4.1-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (239 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m239.5/239.5 kB\u001b[0m \u001b[31m71.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading gitdb-4.0.11-py3-none-any.whl (62 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m62.7/62.7 kB\u001b[0m \u001b[31m23.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading google_api_core-2.15.0-py3-none-any.whl (121 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m122.0/122.0 kB\u001b[0m \u001b[31m32.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading google_cloud_core-2.4.1-py2.py3-none-any.whl (29 kB)\n",
|
||||
"Downloading google_resumable_media-2.7.0-py2.py3-none-any.whl (80 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m80.6/80.6 kB\u001b[0m \u001b[31m22.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading kiwisolver-1.4.5-cp310-cp310-manylinux_2_12_x86_64.manylinux2010_x86_64.whl (1.6 MB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.6/1.6 MB\u001b[0m \u001b[31m102.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading markdown_it_py-3.0.0-py3-none-any.whl (87 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m87.5/87.5 kB\u001b[0m \u001b[31m25.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pyparsing-3.1.1-py3-none-any.whl (103 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m103.1/103.1 kB\u001b[0m \u001b[31m32.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading starlette-0.32.0.post1-py3-none-any.whl (70 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m70.0/70.0 kB\u001b[0m \u001b[31m19.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading wrapt-1.16.0-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl (80 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m80.3/80.3 kB\u001b[0m \u001b[31m30.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading yarl-1.9.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (301 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m301.6/301.6 kB\u001b[0m \u001b[31m80.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading markdown2-2.4.12-py2.py3-none-any.whl (41 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m41.2/41.2 kB\u001b[0m \u001b[31m12.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading googleapis_common_protos-1.62.0-py2.py3-none-any.whl (228 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m228.7/228.7 kB\u001b[0m \u001b[31m57.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading pyasn1-0.5.1-py2.py3-none-any.whl (84 kB)\n",
|
||||
"\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m84.9/84.9 kB\u001b[0m \u001b[31m30.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
|
||||
"\u001b[?25hDownloading smmap-5.0.1-py3-none-any.whl (24 kB)\n",
|
||||
"Building wheels for collected packages: flash-attn, optimum, rouge-score, deepspeed, fire, ffmpy, wavedrom\n",
|
||||
" Building wheel for flash-attn (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for flash-attn: filename=flash_attn-2.3.3-cp310-cp310-linux_x86_64.whl size=57042553 sha256=b1df92cb5bd7657d38b789dd48e907aa3e0bd2715c817eb85f3c4320bb11fb3f\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/e5/e6/fa/941802ec61d1afd320d27160ab1db98e6dba65381f84b76d4a\n",
|
||||
" Building wheel for optimum (pyproject.toml) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for optimum: filename=optimum-1.13.2-py3-none-any.whl size=395599 sha256=ff3a73120e1b6eeeda28f76e3fc8cd4cd826e5d66c869b7848ba150e7af79c62\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/6e/b7/2c/79405d98f0943373d8546daeae25a3d377f7659ca0cbe48699\n",
|
||||
" Building wheel for rouge-score (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for rouge-score: filename=rouge_score-0.1.2-py3-none-any.whl size=24932 sha256=8118ecbbcd3529085e794c803f0ddb182fc6c6d3e8a494103b49a94abf1bec37\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/5f/dd/89/461065a73be61a532ff8599a28e9beef17985c9e9c31e541b4\n",
|
||||
" Building wheel for deepspeed (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for deepspeed: filename=deepspeed-0.12.6-py3-none-any.whl size=1306729 sha256=35c46b6f0275b0d3063522e0af4f3cbd9ec1c310114d8917d87cbe2bf43346e2\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/a3/dc/a2/f585faaed4dec84108916dcc8e8a7c129a216df8202ca32984\n",
|
||||
" Building wheel for fire (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for fire: filename=fire-0.5.0-py2.py3-none-any.whl size=116934 sha256=e76d5185f237f34ec69bb8aa657497bef07408978e4f7efdaef48663bb8cd4ef\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/90/d4/f7/9404e5db0116bd4d43e5666eaa3e70ab53723e1e3ea40c9a95\n",
|
||||
" Building wheel for ffmpy (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for ffmpy: filename=ffmpy-0.3.1-py3-none-any.whl size=5579 sha256=da3b54dc0ac1a825a1a233315970ac80b8b4c53ebd9cb2a2cfdeab118f453a64\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/01/a6/d1/1c0828c304a4283b2c1639a09ad86f83d7c487ef34c6b4a1bf\n",
|
||||
" Building wheel for wavedrom (setup.py) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for wavedrom: filename=wavedrom-2.0.3.post3-py2.py3-none-any.whl size=30052 sha256=7f0cbd15d63ee9c120190bac122ab51bbbfc91ee374bc3c046fadb320816c17e\n",
|
||||
" Stored in directory: /root/.cache/pip/wheels/9c/52/8c/38b454b42f712f325e26f633287484c7dc1ad469e1580c5954\n",
|
||||
"Successfully built flash-attn optimum rouge-score deepspeed fire ffmpy wavedrom\n",
|
||||
"Installing collected packages: sentencepiece, pydub, py-cpuinfo, ninja, nh3, hjson, ffmpy, bitsandbytes, appdirs, addict, xxhash, wrapt, werkzeug, websockets, tzdata, typing-extensions, threadpoolctl, termcolor, tensorboard-data-server, svgwrite, smmap, shortuuid, setproctitle, sentry-sdk, semantic-version, scipy, safetensors, rouge, regex, python-multipart, pyparsing, pynvml, pyasn1, pyarrow-hotfix, pyarrow, protobuf, orjson, oauthlib, multidict, mdurl, markdown2, markdown, llvmlite, kiwisolver, joblib, jmespath, importlib-resources, humanfriendly, hf_transfer, h11, grpcio, google-crc32c, gekko, frozenlist, fonttools, einops, docker-pycreds, dill, cycler, contourpy, colorama, cachetools, async-timeout, art, aioitertools, aiofiles, absl-py, yarl, wavedrom, uvicorn, tiktoken, scikit-learn, rsa, responses, requests-oauthlib, pydantic, pyasn1-modules, pandas, numba, nltk, multiprocess, matplotlib, markdown-it-py, httpcore, googleapis-common-protos, google-resumable-media, gitdb, fire, coloredlogs, botocore, aiosignal, xformers, tokenizers, starlette, rouge-score, rich, httpx, google-auth, GitPython, flash-attn, deepspeed, aiohttp, accelerate, wandb, transformers, gradio-client, google-auth-oauthlib, google-api-core, fastapi, altair, aiobotocore, tensorboard, s3fs, peft, gradio, google-cloud-core, fschat, datasets, bert-score, optimum, google-cloud-storage, evaluate, auto-gptq, gcsfs, axolotl\n",
|
||||
" Attempting uninstall: typing-extensions\n",
|
||||
" Found existing installation: typing_extensions 4.7.1\n",
|
||||
" Uninstalling typing_extensions-4.7.1:\n",
|
||||
" Successfully uninstalled typing_extensions-4.7.1\n",
|
||||
" Running setup.py develop for axolotl\n",
|
||||
"Successfully installed GitPython-3.1.40 absl-py-2.0.0 accelerate-0.24.1 addict-2.4.0 aiobotocore-2.7.0 aiofiles-23.2.1 aiohttp-3.9.1 aioitertools-0.11.0 aiosignal-1.3.1 altair-5.2.0 appdirs-1.4.4 art-6.1 async-timeout-4.0.3 auto-gptq-0.5.1 axolotl-0.3.0 bert-score-0.3.13 bitsandbytes-0.41.3.post2 botocore-1.31.64 cachetools-5.3.2 colorama-0.4.6 coloredlogs-15.0.1 contourpy-1.2.0 cycler-0.12.1 datasets-2.16.0 deepspeed-0.12.6 dill-0.3.7 docker-pycreds-0.4.0 einops-0.7.0 evaluate-0.4.0 fastapi-0.108.0 ffmpy-0.3.1 fire-0.5.0 flash-attn-2.3.3 fonttools-4.47.0 frozenlist-1.4.1 fschat-0.2.34 gcsfs-2023.10.0 gekko-1.0.6 gitdb-4.0.11 google-api-core-2.15.0 google-auth-2.25.2 google-auth-oauthlib-1.2.0 google-cloud-core-2.4.1 google-cloud-storage-2.14.0 google-crc32c-1.5.0 google-resumable-media-2.7.0 googleapis-common-protos-1.62.0 gradio-3.50.2 gradio-client-0.6.1 grpcio-1.60.0 h11-0.14.0 hf_transfer-0.1.4 hjson-3.1.0 httpcore-1.0.2 httpx-0.26.0 humanfriendly-10.0 importlib-resources-6.1.1 jmespath-1.0.1 joblib-1.3.2 kiwisolver-1.4.5 llvmlite-0.41.1 markdown-3.5.1 markdown-it-py-3.0.0 markdown2-2.4.12 matplotlib-3.8.2 mdurl-0.1.2 multidict-6.0.4 multiprocess-0.70.15 nh3-0.2.15 ninja-1.11.1.1 nltk-3.8.1 numba-0.58.1 oauthlib-3.2.2 optimum-1.13.2 orjson-3.9.10 pandas-2.1.4 peft-0.6.0 protobuf-4.23.4 py-cpuinfo-9.0.0 pyarrow-14.0.2 pyarrow-hotfix-0.6 pyasn1-0.5.1 pyasn1-modules-0.3.0 pydantic-1.10.13 pydub-0.25.1 pynvml-11.5.0 pyparsing-3.1.1 python-multipart-0.0.6 regex-2023.12.25 requests-oauthlib-1.3.1 responses-0.18.0 rich-13.7.0 rouge-1.0.1 rouge-score-0.1.2 rsa-4.9 s3fs-2023.10.0 safetensors-0.4.1 scikit-learn-1.2.2 scipy-1.11.4 semantic-version-2.10.0 sentencepiece-0.1.99 sentry-sdk-1.39.1 setproctitle-1.3.3 shortuuid-1.0.11 smmap-5.0.1 starlette-0.32.0.post1 svgwrite-1.4.3 tensorboard-2.15.1 tensorboard-data-server-0.7.2 termcolor-2.4.0 threadpoolctl-3.2.0 tiktoken-0.5.2 tokenizers-0.15.0 transformers-4.36.2 typing-extensions-4.8.0 tzdata-2023.3 uvicorn-0.25.0 wandb-0.16.1 wavedrom-2.0.3.post3 websockets-11.0.3 werkzeug-3.0.1 wrapt-1.16.0 xformers-0.0.23 xxhash-3.4.1 yarl-1.9.4\n",
|
||||
"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
|
||||
"\u001b[0mCollecting git+https://github.com/huggingface/peft.git\n",
|
||||
" Cloning https://github.com/huggingface/peft.git to /tmp/pip-req-build-hka8xgk2\n",
|
||||
" Running command git clone --filter=blob:none --quiet https://github.com/huggingface/peft.git /tmp/pip-req-build-hka8xgk2\n",
|
||||
" Resolved https://github.com/huggingface/peft.git to commit cf04d0353f0343cbf66627228c4495f51669af34\n",
|
||||
" Installing build dependencies ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Getting requirements to build wheel ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Preparing metadata (pyproject.toml) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25hRequirement already satisfied: numpy>=1.17 in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (1.26.0)\n",
|
||||
"Requirement already satisfied: packaging>=20.0 in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (23.1)\n",
|
||||
"Requirement already satisfied: psutil in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (5.9.0)\n",
|
||||
"Requirement already satisfied: pyyaml in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (6.0.1)\n",
|
||||
"Requirement already satisfied: torch>=1.13.0 in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (2.1.1)\n",
|
||||
"Requirement already satisfied: transformers in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (4.36.2)\n",
|
||||
"Requirement already satisfied: tqdm in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (4.65.0)\n",
|
||||
"Requirement already satisfied: accelerate>=0.21.0 in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (0.24.1)\n",
|
||||
"Requirement already satisfied: safetensors in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (0.4.1)\n",
|
||||
"Requirement already satisfied: huggingface-hub>=0.17.0 in /opt/conda/lib/python3.10/site-packages (from peft==0.7.2.dev0) (0.20.1)\n",
|
||||
"Requirement already satisfied: filelock in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.17.0->peft==0.7.2.dev0) (3.9.0)\n",
|
||||
"Requirement already satisfied: fsspec>=2023.5.0 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.17.0->peft==0.7.2.dev0) (2023.10.0)\n",
|
||||
"Requirement already satisfied: requests in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.17.0->peft==0.7.2.dev0) (2.31.0)\n",
|
||||
"Requirement already satisfied: typing-extensions>=3.7.4.3 in /opt/conda/lib/python3.10/site-packages (from huggingface-hub>=0.17.0->peft==0.7.2.dev0) (4.8.0)\n",
|
||||
"Requirement already satisfied: sympy in /opt/conda/lib/python3.10/site-packages (from torch>=1.13.0->peft==0.7.2.dev0) (1.11.1)\n",
|
||||
"Requirement already satisfied: networkx in /opt/conda/lib/python3.10/site-packages (from torch>=1.13.0->peft==0.7.2.dev0) (3.1)\n",
|
||||
"Requirement already satisfied: jinja2 in /opt/conda/lib/python3.10/site-packages (from torch>=1.13.0->peft==0.7.2.dev0) (3.1.2)\n",
|
||||
"Requirement already satisfied: regex!=2019.12.17 in /opt/conda/lib/python3.10/site-packages (from transformers->peft==0.7.2.dev0) (2023.12.25)\n",
|
||||
"Requirement already satisfied: tokenizers<0.19,>=0.14 in /opt/conda/lib/python3.10/site-packages (from transformers->peft==0.7.2.dev0) (0.15.0)\n",
|
||||
"Requirement already satisfied: MarkupSafe>=2.0 in /opt/conda/lib/python3.10/site-packages (from jinja2->torch>=1.13.0->peft==0.7.2.dev0) (2.1.1)\n",
|
||||
"Requirement already satisfied: charset-normalizer<4,>=2 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.17.0->peft==0.7.2.dev0) (2.0.4)\n",
|
||||
"Requirement already satisfied: idna<4,>=2.5 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.17.0->peft==0.7.2.dev0) (3.4)\n",
|
||||
"Requirement already satisfied: urllib3<3,>=1.21.1 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.17.0->peft==0.7.2.dev0) (1.26.18)\n",
|
||||
"Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.10/site-packages (from requests->huggingface-hub>=0.17.0->peft==0.7.2.dev0) (2023.7.22)\n",
|
||||
"Requirement already satisfied: mpmath>=0.19 in /opt/conda/lib/python3.10/site-packages (from sympy->torch>=1.13.0->peft==0.7.2.dev0) (1.3.0)\n",
|
||||
"Building wheels for collected packages: peft\n",
|
||||
" Building wheel for peft (pyproject.toml) ... \u001b[?25ldone\n",
|
||||
"\u001b[?25h Created wheel for peft: filename=peft-0.7.2.dev0-py3-none-any.whl size=169456 sha256=4c70d23e759fa6abb3827fb2f3a8683be3b24d78777d0f403bbc2c0548e5dd4b\n",
|
||||
" Stored in directory: /tmp/pip-ephem-wheel-cache-my5ncou6/wheels/d7/c7/de/1368fac8590e1b103ddc2ec2a28ad51d83aded1a3830e8a087\n",
|
||||
"Successfully built peft\n",
|
||||
"Installing collected packages: peft\n",
|
||||
" Attempting uninstall: peft\n",
|
||||
" Found existing installation: peft 0.6.0\n",
|
||||
" Uninstalling peft-0.6.0:\n",
|
||||
" Successfully uninstalled peft-0.6.0\n",
|
||||
"\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
|
||||
"axolotl 0.3.0 requires peft==0.6.0, but you have peft 0.7.2.dev0 which is incompatible.\u001b[0m\u001b[31m\n",
|
||||
"\u001b[0mSuccessfully installed peft-0.7.2.dev0\n",
|
||||
"\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\u001b[33m\n",
|
||||
"\u001b[0m"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"#instaling what is needed inside axolotl file\n",
|
||||
"!pip install packaging\n",
|
||||
"!pip install -e '.[flash-attn,deepspeed]'\n",
|
||||
"!pip install -U git+https://github.com/huggingface/peft.git"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 16,
|
||||
"id": "82d1a380-1e87-48fe-89fe-25331326014d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The following values were not passed to `accelerate launch` and had defaults used instead:\n",
|
||||
"\t`--num_processes` was set to a value of `3`\n",
|
||||
"\t\tMore than one GPU was found, enabling multi-GPU training.\n",
|
||||
"\t\tIf this was unintended please pass in `--num_processes=1`.\n",
|
||||
"\t`--num_machines` was set to a value of `1`\n",
|
||||
"\t`--mixed_precision` was set to a value of `'no'`\n",
|
||||
"\t`--dynamo_backend` was set to a value of `'no'`\n",
|
||||
"To avoid this warning pass in values for each of the problematic parameters or run `accelerate config`.\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\n",
|
||||
" warnings.warn(\n",
|
||||
"[2023-12-28 15:44:09,979] [INFO] [datasets.<module>:58] [PID:2814] PyTorch version 2.1.1 available.\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/transformers/deepspeed.py:23: FutureWarning: transformers.deepspeed module is deprecated and will be removed in a future version. Please import deepspeed modules directly from transformers.integrations\n",
|
||||
" warnings.warn(\n",
|
||||
"[2023-12-28 15:44:10,011] [INFO] [datasets.<module>:58] [PID:2812] PyTorch version 2.1.1 available.\n",
|
||||
"[2023-12-28 15:44:10,013] [INFO] [datasets.<module>:58] [PID:2813] PyTorch version 2.1.1 available.\n",
|
||||
"[2023-12-28 15:44:10,805] [INFO] [axolotl.normalize_config:150] [PID:2814] [RANK:2] GPU memory usage baseline: 0.000GB (+0.317GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:10,830] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)\n",
|
||||
"[2023-12-28 15:44:10,842] [INFO] [axolotl.normalize_config:150] [PID:2813] [RANK:1] GPU memory usage baseline: 0.000GB (+0.317GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:10,865] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)\n",
|
||||
"[2023-12-28 15:44:10,869] [INFO] [axolotl.normalize_config:150] [PID:2812] [RANK:0] GPU memory usage baseline: 0.000GB (+0.351GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:10,887] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to cuda (auto detect)\n",
|
||||
"[2023-12-28 15:44:10,961] [INFO] [comm.py:637:init_distributed] cdb=None\n",
|
||||
"[2023-12-28 15:44:10,994] [INFO] [comm.py:637:init_distributed] cdb=None\n",
|
||||
"[2023-12-28 15:44:11,015] [INFO] [comm.py:637:init_distributed] cdb=None\n",
|
||||
"[2023-12-28 15:44:11,015] [INFO] [comm.py:668:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl\n",
|
||||
" dP dP dP \n",
|
||||
" 88 88 88 \n",
|
||||
" .d8888b. dP. .dP .d8888b. 88 .d8888b. d8888P 88 \n",
|
||||
" 88' `88 `8bd8' 88' `88 88 88' `88 88 88 \n",
|
||||
" 88. .88 .d88b. 88. .88 88 88. .88 88 88 \n",
|
||||
" `88888P8 dP' `dP `88888P' dP `88888P' dP dP \n",
|
||||
" \n",
|
||||
" \n",
|
||||
"\n",
|
||||
"[2023-12-28 15:44:11,412] [DEBUG] [axolotl.load_tokenizer:184] [PID:2812] [RANK:0] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,412] [DEBUG] [axolotl.load_tokenizer:185] [PID:2812] [RANK:0] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,412] [DEBUG] [axolotl.load_tokenizer:186] [PID:2812] [RANK:0] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,412] [DEBUG] [axolotl.load_tokenizer:187] [PID:2812] [RANK:0] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,413] [INFO] [axolotl.load_tokenized_prepared_datasets:143] [PID:2812] [RANK:0] Loading prepared dataset from disk at tilemachos/GF_new.json/1adc45d2edc1e98ce657814412c6593c...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,415] [INFO] [axolotl.load_tokenized_prepared_datasets:145] [PID:2812] [RANK:0] Prepared dataset loaded from disk...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,432] [DEBUG] [axolotl.load_tokenizer:184] [PID:2814] [RANK:2] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,432] [DEBUG] [axolotl.load_tokenizer:185] [PID:2814] [RANK:2] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,432] [DEBUG] [axolotl.load_tokenizer:186] [PID:2814] [RANK:2] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,432] [DEBUG] [axolotl.load_tokenizer:187] [PID:2814] [RANK:2] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,530] [DEBUG] [axolotl.load_tokenizer:184] [PID:2813] [RANK:1] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,531] [DEBUG] [axolotl.load_tokenizer:185] [PID:2813] [RANK:1] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,531] [DEBUG] [axolotl.load_tokenizer:186] [PID:2813] [RANK:1] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:11,531] [DEBUG] [axolotl.load_tokenizer:187] [PID:2813] [RANK:1] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,158] [INFO] [axolotl.load_tokenized_prepared_datasets:143] [PID:2813] [RANK:1] Loading prepared dataset from disk at tilemachos/GF_new.json/1adc45d2edc1e98ce657814412c6593c...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,158] [INFO] [axolotl.load_tokenized_prepared_datasets:143] [PID:2814] [RANK:2] Loading prepared dataset from disk at tilemachos/GF_new.json/1adc45d2edc1e98ce657814412c6593c...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,160] [INFO] [axolotl.load_tokenized_prepared_datasets:145] [PID:2813] [RANK:1] Prepared dataset loaded from disk...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,161] [INFO] [axolotl.load_tokenized_prepared_datasets:145] [PID:2814] [RANK:2] Prepared dataset loaded from disk...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,236] [DEBUG] [axolotl.log:60] [PID:2812] [RANK:0] total_num_tokens: 28120\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,238] [DEBUG] [axolotl.log:60] [PID:2812] [RANK:0] `total_supervised_tokens: 7990`\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,238] [DEBUG] [axolotl.log:60] [PID:2812] [RANK:0] total_num_steps: 6\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,242] [DEBUG] [axolotl.train.log:60] [PID:2812] [RANK:0] loading tokenizer... mistralai/Mistral-7B-v0.1\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,518] [DEBUG] [axolotl.load_tokenizer:184] [PID:2812] [RANK:0] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,518] [DEBUG] [axolotl.load_tokenizer:185] [PID:2812] [RANK:0] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,518] [DEBUG] [axolotl.load_tokenizer:186] [PID:2812] [RANK:0] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,518] [DEBUG] [axolotl.load_tokenizer:187] [PID:2812] [RANK:0] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,518] [DEBUG] [axolotl.train.log:60] [PID:2812] [RANK:0] loading model and peft_config...\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,589] [DEBUG] [axolotl.load_tokenizer:184] [PID:2814] [RANK:2] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,589] [DEBUG] [axolotl.load_tokenizer:185] [PID:2814] [RANK:2] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,589] [DEBUG] [axolotl.load_tokenizer:186] [PID:2814] [RANK:2] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,589] [DEBUG] [axolotl.load_tokenizer:187] [PID:2814] [RANK:2] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,599] [DEBUG] [axolotl.load_tokenizer:184] [PID:2813] [RANK:1] EOS: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,599] [DEBUG] [axolotl.load_tokenizer:185] [PID:2813] [RANK:1] BOS: 1 / <s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,599] [DEBUG] [axolotl.load_tokenizer:186] [PID:2813] [RANK:1] PAD: 2 / </s>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:12,599] [DEBUG] [axolotl.load_tokenizer:187] [PID:2813] [RANK:1] UNK: 0 / <unk>\u001b[39m\n",
|
||||
"[2023-12-28 15:44:13,049] [INFO] [partition_parameters.py:348:__exit__] finished initializing model - num_params = 291, num_elems = 7.24B\n",
|
||||
"Loading checkpoint shards: 100%|██████████████████| 2/2 [00:11<00:00, 5.81s/it]\n",
|
||||
"Loading checkpoint shards: 100%|██████████████████| 2/2 [00:11<00:00, 5.98s/it]\n",
|
||||
"[2023-12-28 15:44:25,395] [INFO] [axolotl.load_model:503] [PID:2813] [RANK:1] GPU memory usage after model load: 7.576GB (+0.524GB cache, +0.708GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:25,399] [INFO] [axolotl.load_model:526] [PID:2813] [RANK:1] converting PEFT model w/ prepare_model_for_kbit_training\u001b[39m\n",
|
||||
"[2023-12-28 15:44:25,403] [INFO] [axolotl.load_model:538] [PID:2813] [RANK:1] converting modules to torch.bfloat16 for flash attention\u001b[39m\n",
|
||||
"trainable params: 3,407,872 || all params: 7,245,139,968 || trainable%: 0.04703666202518836\n",
|
||||
"[2023-12-28 15:44:25,480] [INFO] [axolotl.load_model:568] [PID:2813] [RANK:1] GPU memory usage after adapters: 7.589GB (+1.501GB cache, +0.708GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:25,572] [INFO] [axolotl.load_model:503] [PID:2814] [RANK:2] GPU memory usage after model load: 7.576GB (+0.410GB cache, +0.708GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:25,576] [INFO] [axolotl.load_model:526] [PID:2814] [RANK:2] converting PEFT model w/ prepare_model_for_kbit_training\u001b[39m\n",
|
||||
"[2023-12-28 15:44:25,580] [INFO] [axolotl.load_model:538] [PID:2814] [RANK:2] converting modules to torch.bfloat16 for flash attention\u001b[39m\n",
|
||||
"trainable params: 3,407,872 || all params: 7,245,139,968 || trainable%: 0.04703666202518836\n",
|
||||
"[2023-12-28 15:44:25,660] [INFO] [axolotl.load_model:568] [PID:2814] [RANK:2] GPU memory usage after adapters: 7.589GB (+1.388GB cache, +0.708GB misc)\u001b[39m\n",
|
||||
"Loading checkpoint shards: 100%|██████████████████| 2/2 [00:12<00:00, 6.30s/it]\n",
|
||||
"[2023-12-28 15:44:26,170] [INFO] [axolotl.load_model:503] [PID:2812] [RANK:0] GPU memory usage after model load: 7.576GB (+0.776GB cache, +0.741GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:26,177] [INFO] [axolotl.load_model:526] [PID:2812] [RANK:0] converting PEFT model w/ prepare_model_for_kbit_training\u001b[39m\n",
|
||||
"[2023-12-28 15:44:26,181] [INFO] [axolotl.load_model:538] [PID:2812] [RANK:0] converting modules to torch.bfloat16 for flash attention\u001b[39m\n",
|
||||
"trainable params: 3,407,872 || all params: 7,245,139,968 || trainable%: 0.04703666202518836\n",
|
||||
"[2023-12-28 15:44:26,259] [INFO] [axolotl.load_model:568] [PID:2812] [RANK:0] GPU memory usage after adapters: 7.589GB (+1.753GB cache, +0.741GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:44:26,293] [INFO] [axolotl.train.log:60] [PID:2812] [RANK:0] Pre-saving adapter config to ./out\u001b[39m\n",
|
||||
"[2023-12-28 15:44:26,296] [INFO] [axolotl.train.log:60] [PID:2812] [RANK:0] Starting trainer...\u001b[39m\n",
|
||||
"Using /root/.cache/torch_extensions/py310_cu121 as PyTorch extensions root...\n",
|
||||
"Using /root/.cache/torch_extensions/py310_cu121 as PyTorch extensions root...\n",
|
||||
"Using /root/.cache/torch_extensions/py310_cu121 as PyTorch extensions root...\n",
|
||||
"Detected CUDA files, patching ldflags\n",
|
||||
"Emitting ninja build file /root/.cache/torch_extensions/py310_cu121/fused_adam/build.ninja...\n",
|
||||
"Building extension module fused_adam...\n",
|
||||
"Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)\n",
|
||||
"ninja: no work to do.\n",
|
||||
"Loading extension module fused_adam...\n",
|
||||
"Time to load fused_adam op: 0.05891108512878418 seconds\n",
|
||||
"Loading extension module fused_adam...\n",
|
||||
"Time to load fused_adam op: 0.10173463821411133 seconds\n",
|
||||
"Loading extension module fused_adam...\n",
|
||||
"Time to load fused_adam op: 0.10152459144592285 seconds\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/deepspeed/ops/adam/fused_adam.py:96: UserWarning: The torch.cuda.*DtypeTensor constructors are no longer recommended. It's best to use methods such as torch.tensor(data, dtype=*, device='cuda') to create tensors. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/tensor/python_tensor.cpp:83.)\n",
|
||||
" self._dummy_overflow_buf = get_accelerator().IntTensor([0])\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/deepspeed/ops/adam/fused_adam.py:96: UserWarning: The torch.cuda.*DtypeTensor constructors are no longer recommended. It's best to use methods such as torch.tensor(data, dtype=*, device='cuda') to create tensors. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/tensor/python_tensor.cpp:83.)\n",
|
||||
" self._dummy_overflow_buf = get_accelerator().IntTensor([0])\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/deepspeed/ops/adam/fused_adam.py:96: UserWarning: The torch.cuda.*DtypeTensor constructors are no longer recommended. It's best to use methods such as torch.tensor(data, dtype=*, device='cuda') to create tensors. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/tensor/python_tensor.cpp:83.)\n",
|
||||
" self._dummy_overflow_buf = get_accelerator().IntTensor([0])\n",
|
||||
"Parameter Offload: Total persistent parameters: 3674112 in 193 params\n",
|
||||
" 0%| | 0/17 [00:00<?, ?it/s]/opt/conda/lib/python3.10/site-packages/torch/utils/checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants.\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/torch/utils/checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants.\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/torch/utils/checkpoint.py:429: UserWarning: torch.utils.checkpoint: please pass in use_reentrant=True or use_reentrant=False explicitly. The default value of use_reentrant will be updated to be False in the future. To maintain current behavior, pass use_reentrant=True. It is recommended that you use use_reentrant=False. Refer to docs for more details on the differences between the two variants.\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/bitsandbytes/autograd/_functions.py:322: UserWarning: MatMul8bitLt: inputs will be cast from torch.bfloat16 to float16 during quantization\n",
|
||||
" warnings.warn(f\"MatMul8bitLt: inputs will be cast from {A.dtype} to float16 during quantization\")\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/bitsandbytes/autograd/_functions.py:322: UserWarning: MatMul8bitLt: inputs will be cast from torch.bfloat16 to float16 during quantization\n",
|
||||
" warnings.warn(f\"MatMul8bitLt: inputs will be cast from {A.dtype} to float16 during quantization\")\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/bitsandbytes/autograd/_functions.py:322: UserWarning: MatMul8bitLt: inputs will be cast from torch.bfloat16 to float16 during quantization\n",
|
||||
" warnings.warn(f\"MatMul8bitLt: inputs will be cast from {A.dtype} to float16 during quantization\")\n",
|
||||
"{'loss': 2.0448, 'learning_rate': 2e-05, 'epoch': 0.06} \n",
|
||||
" 6%|██▌ | 1/17 [00:28<07:32, 28.30s/it]\n",
|
||||
" 0%| | 0/3 [00:00<?, ?it/s]\u001b[A\n",
|
||||
" 67%|██████████████████████████████ | 2/3 [00:03<00:01, 1.85s/it]\u001b[A\n",
|
||||
" \u001b[A\n",
|
||||
"\u001b[A{'eval_loss': 1.9694719314575195, 'eval_runtime': 11.391, 'eval_samples_per_second': 1.492, 'eval_steps_per_second': 0.263, 'epoch': 0.06}\n",
|
||||
" 6%|██▌ | 1/17 [00:39<07:32, 28.30s/it]\n",
|
||||
"100%|█████████████████████████████████████████████| 3/3 [00:07<00:00, 2.65s/it]\u001b[A\n",
|
||||
" \u001b[A[2023-12-28 15:45:35,358] [INFO] [axolotl.callbacks.on_step_end:122] [PID:2812] [RANK:0] GPU memory usage while training: 12.210GB (+4.259GB cache, +0.776GB misc)\u001b[39m\n",
|
||||
" 12%|█████▏ | 2/17 [01:04<08:18, 33.20s/it][2023-12-28 15:45:35,358] [INFO] [axolotl.callbacks.on_step_end:122] [PID:2814] [RANK:2] GPU memory usage while training: 12.269GB (+4.522GB cache, +0.743GB misc)\u001b[39m\n",
|
||||
"[2023-12-28 15:45:35,358] [INFO] [axolotl.callbacks.on_step_end:122] [PID:2813] [RANK:1] GPU memory usage while training: 12.283GB (+4.493GB cache, +0.743GB misc)\u001b[39m\n",
|
||||
"{'loss': 2.0022, 'learning_rate': 4e-05, 'epoch': 0.12} \n",
|
||||
"{'loss': 2.1054, 'learning_rate': 6e-05, 'epoch': 0.17} \n",
|
||||
"{'loss': 1.9004, 'learning_rate': 8e-05, 'epoch': 0.23} \n",
|
||||
"{'loss': 1.8794, 'learning_rate': 0.0001, 'epoch': 0.29} \n",
|
||||
" 29%|████████████▉ | 5/17 [02:20<05:23, 26.92s/it]\n",
|
||||
" 0%| | 0/3 [00:00<?, ?it/s]\u001b[A\n",
|
||||
" 67%|██████████████████████████████ | 2/3 [00:03<00:01, 1.88s/it]\u001b[A\n",
|
||||
" \u001b[A\n",
|
||||
"\u001b[A{'eval_loss': 1.7912336587905884, 'eval_runtime': 11.3106, 'eval_samples_per_second': 1.503, 'eval_steps_per_second': 0.265, 'epoch': 0.29}\n",
|
||||
" 29%|████████████▉ | 5/17 [02:32<05:23, 26.92s/it]\n",
|
||||
"100%|█████████████████████████████████████████████| 3/3 [00:07<00:00, 2.67s/it]\u001b[A\n",
|
||||
"{'loss': 1.7871, 'learning_rate': 0.00012, 'epoch': 0.35} \u001b[A\n",
|
||||
"{'loss': 1.7758, 'learning_rate': 0.00014, 'epoch': 0.4} \n",
|
||||
"{'loss': 1.4645, 'learning_rate': 0.00016, 'epoch': 0.46} \n",
|
||||
"{'loss': 1.4009, 'learning_rate': 0.00018, 'epoch': 0.52} \n",
|
||||
"{'loss': 1.3927, 'learning_rate': 0.0002, 'epoch': 0.58} \n",
|
||||
" 59%|█████████████████████████▎ | 10/17 [04:38<03:04, 26.33s/it]\n",
|
||||
" 0%| | 0/3 [00:00<?, ?it/s]\u001b[A\n",
|
||||
" 67%|██████████████████████████████ | 2/3 [00:03<00:01, 1.89s/it]\u001b[A\n",
|
||||
" \u001b[A\n",
|
||||
"\u001b[A{'eval_loss': 1.1426481008529663, 'eval_runtime': 11.3344, 'eval_samples_per_second': 1.5, 'eval_steps_per_second': 0.265, 'epoch': 0.58}\n",
|
||||
" 59%|█████████████████████████▎ | 10/17 [04:49<03:04, 26.33s/it]\n",
|
||||
"100%|█████████████████████████████████████████████| 3/3 [00:07<00:00, 2.68s/it]\u001b[A\n",
|
||||
"{'loss': 1.0122, 'learning_rate': 0.0001900968867902419, 'epoch': 0.63} \u001b[A\n",
|
||||
"{'loss': 1.0019, 'learning_rate': 0.00016234898018587337, 'epoch': 0.69} \n",
|
||||
"{'loss': 0.8976, 'learning_rate': 0.00012225209339563145, 'epoch': 0.75} \n",
|
||||
"{'loss': 0.9301, 'learning_rate': 7.774790660436858e-05, 'epoch': 0.81} \n",
|
||||
"{'loss': 0.8595, 'learning_rate': 3.7651019814126654e-05, 'epoch': 0.87} \n",
|
||||
" 88%|█████████████████████████████████████▉ | 15/17 [06:55<00:52, 26.17s/it]\n",
|
||||
" 0%| | 0/3 [00:00<?, ?it/s]\u001b[A\n",
|
||||
" 67%|██████████████████████████████ | 2/3 [00:03<00:01, 1.88s/it]\u001b[A\n",
|
||||
" \u001b[A\n",
|
||||
"\u001b[A{'eval_loss': 0.8175248503684998, 'eval_runtime': 11.2932, 'eval_samples_per_second': 1.505, 'eval_steps_per_second': 0.266, 'epoch': 0.87}\n",
|
||||
" 88%|█████████████████████████████████████▉ | 15/17 [07:06<00:52, 26.17s/it]\n",
|
||||
"100%|█████████████████████████████████████████████| 3/3 [00:07<00:00, 2.67s/it]\u001b[A\n",
|
||||
"{'loss': 0.7931, 'learning_rate': 9.903113209758096e-06, 'epoch': 0.92} \u001b[A\n",
|
||||
"{'loss': 0.6909, 'learning_rate': 0.0, 'epoch': 0.98} \n",
|
||||
"100%|███████████████████████████████████████████| 17/17 [07:56<00:00, 28.03s/it]/opt/conda/lib/python3.10/site-packages/torch/nn/modules/module.py:1879: UserWarning: Positional args are being deprecated, use kwargs instead. Refer to https://pytorch.org/docs/master/generated/torch.nn.Module.html#torch.nn.Module.state_dict for details.\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/torch/nn/modules/module.py:1879: UserWarning: Positional args are being deprecated, use kwargs instead. Refer to https://pytorch.org/docs/master/generated/torch.nn.Module.html#torch.nn.Module.state_dict for details.\n",
|
||||
" warnings.warn(\n",
|
||||
"/opt/conda/lib/python3.10/site-packages/torch/nn/modules/module.py:1879: UserWarning: Positional args are being deprecated, use kwargs instead. Refer to https://pytorch.org/docs/master/generated/torch.nn.Module.html#torch.nn.Module.state_dict for details.\n",
|
||||
" warnings.warn(\n",
|
||||
"{'train_runtime': 489.0649, 'train_samples_per_second': 0.63, 'train_steps_per_second': 0.035, 'train_loss': 1.408153467318591, 'epoch': 0.98}\n",
|
||||
"100%|███████████████████████████████████████████| 17/17 [08:09<00:00, 28.77s/it]\n",
|
||||
"[2023-12-28 15:52:39,488] [INFO] [axolotl.train.log:60] [PID:2812] [RANK:0] Training Completed!!! Saving pre-trained model to ./out\u001b[39m\n",
|
||||
"\u001b[0m\u001b[0m\u001b[0m"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"\"\"\"\n",
|
||||
"Training using the config.yml file and using deepspeed:zero3_bf16 the most aggressive optimization out of zero1,zero2,zero3 stages which partitions \n",
|
||||
"not only optimizer states but also gradients and parameters across GPUs. The bf16 indicate mixed precision training using bfloat16.\n",
|
||||
"For more information read axolotl's readme\n",
|
||||
"\"\"\"\n",
|
||||
"!accelerate launch -m axolotl.cli.train /folder/config.yml --deepspeed deepspeed_configs/zero3_bf16.json"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.10.13"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@@ -1,10 +0,0 @@
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: Who is the Founder of Apple\""}, {"from": "gpt", "value": "\"<Chatbot>: The founder of Apple is Steve Jobs\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: What is the capital of France?\""}, {"from": "gpt", "value": "\"<Chatbot>: The capital of France is Paris.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: How far is the Moon from Earth?\""}, {"from": "gpt", "value": "\"<Chatbot>: The Moon is approximately 384,400 kilometers from Earth.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: What is the tallest mountain in the world?\""}, {"from": "gpt", "value": "\"<Chatbot>: The tallest mountain in the world is Mount Everest.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: Who wrote Romeo and Juliet?\""}, {"from": "gpt", "value": "\"<Chatbot>: Romeo and Juliet was written by William Shakespeare.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: What is the boiling point of water?\""}, {"from": "gpt", "value": "\"<Chatbot>: The boiling point of water is 100 degrees Celsius.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: When was the first man on the moon?\""}, {"from": "gpt", "value": "\"<Chatbot>: The first man landed on the moon in 1969.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: What is the largest ocean?\""}, {"from": "gpt", "value": "\"<Chatbot>: The largest ocean is the Pacific Ocean.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: Who invented the telephone?\""}, {"from": "gpt", "value": "\"<Chatbot>: The telephone was invented by Alexander Graham Bell.\""}]}
|
||||
{"conversations": [{"from": "Customer", "value": "\"<Customer>: What is the formula for water?\""}, {"from": "gpt", "value": "\"<Chatbot>: The chemical formula for water is H2O.\""}]}
|
||||
@@ -56,6 +56,3 @@ weight_decay: 0.0
|
||||
fsdp:
|
||||
fsdp_config:
|
||||
special_tokens:
|
||||
bos_token: "<s>"
|
||||
eos_token: "</s>"
|
||||
unk_token: "<unk>"
|
||||
|
||||
@@ -1,4 +1,3 @@
|
||||
#Mistral-7b
|
||||
base_model: mistralai/Mistral-7B-v0.1
|
||||
model_type: MistralForCausalLM
|
||||
tokenizer_type: LlamaTokenizer
|
||||
@@ -8,26 +7,32 @@ load_in_4bit: false
|
||||
strict: false
|
||||
|
||||
datasets:
|
||||
- path: tilemachos/Demo-Dataset #Path to json dataset file in huggingface
|
||||
#for type,conversation arguments read axolotl readme and pick what is suited for your project, I wanted a chatbot and put sharegpt and chatml
|
||||
type: sharegpt
|
||||
conversation: chatml
|
||||
dataset_prepared_path: tilemachos/Demo-Dataset #Path to json dataset file in huggingface
|
||||
val_set_size: 0.05
|
||||
output_dir: ./out
|
||||
- path: mhenrichsen/alpaca_2k_test
|
||||
type: alpaca
|
||||
dataset_prepared_path: last_run_prepared
|
||||
val_set_size: 0.1
|
||||
output_dir: ./lora-out
|
||||
|
||||
#using lora for lower cost
|
||||
adapter: lora
|
||||
lora_r: 8
|
||||
lora_model_dir:
|
||||
|
||||
sequence_len: 8192
|
||||
sample_packing: true
|
||||
pad_to_sequence_len: true
|
||||
|
||||
lora_r: 32
|
||||
lora_alpha: 16
|
||||
lora_dropout: 0.05
|
||||
lora_target_linear: true
|
||||
lora_fan_in_fan_out:
|
||||
lora_target_modules:
|
||||
- gate_proj
|
||||
- down_proj
|
||||
- up_proj
|
||||
- q_proj
|
||||
- v_proj
|
||||
|
||||
sequence_len: 512
|
||||
sample_packing: false
|
||||
pad_to_sequence_len: true
|
||||
- k_proj
|
||||
- o_proj
|
||||
|
||||
wandb_project:
|
||||
wandb_entity:
|
||||
@@ -35,18 +40,17 @@ wandb_watch:
|
||||
wandb_name:
|
||||
wandb_log_model:
|
||||
|
||||
#only 2 epochs because of small dataset
|
||||
gradient_accumulation_steps: 3
|
||||
gradient_accumulation_steps: 4
|
||||
micro_batch_size: 2
|
||||
num_epochs: 2
|
||||
num_epochs: 1
|
||||
optimizer: adamw_bnb_8bit
|
||||
lr_scheduler: cosine
|
||||
learning_rate: 0.0002
|
||||
|
||||
train_on_inputs: false
|
||||
group_by_length: false
|
||||
bf16: true
|
||||
fp16: false
|
||||
bf16: auto
|
||||
fp16:
|
||||
tf32: false
|
||||
|
||||
gradient_checkpointing: true
|
||||
@@ -57,18 +61,17 @@ logging_steps: 1
|
||||
xformers_attention:
|
||||
flash_attention: true
|
||||
|
||||
loss_watchdog_threshold: 5.0
|
||||
loss_watchdog_patience: 3
|
||||
|
||||
warmup_steps: 10
|
||||
evals_per_epoch: 4
|
||||
eval_table_size:
|
||||
eval_max_new_tokens: 128
|
||||
saves_per_epoch: 1
|
||||
debug:
|
||||
#default deepspeed, can use more aggresive if needed like zero2, zero3
|
||||
deepspeed: deepspeed_configs/zero1.json
|
||||
deepspeed:
|
||||
weight_decay: 0.0
|
||||
fsdp:
|
||||
fsdp_config:
|
||||
special_tokens:
|
||||
bos_token: "<s>"
|
||||
eos_token: "</s>"
|
||||
unk_token: "<unk>"
|
||||
@@ -75,6 +75,3 @@ weight_decay: 0.0
|
||||
fsdp:
|
||||
fsdp_config:
|
||||
special_tokens:
|
||||
bos_token: "<s>"
|
||||
eos_token: "</s>"
|
||||
unk_token: "<unk>"
|
||||
|
||||
BIN
favicon.jpg
Normal file
BIN
favicon.jpg
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 4.5 KiB |
19
index.qmd
Normal file
19
index.qmd
Normal file
@@ -0,0 +1,19 @@
|
||||
|
||||
|
||||
```{python}
|
||||
#|output: asis
|
||||
#|echo: false
|
||||
|
||||
# This cell steals the README as the home page for now, but excludes the table of contents (quarto adds its own)
|
||||
import re
|
||||
pattern = re.compile(
|
||||
r"<table>\s*<tr>\s*<td>\s*## Table of Contents.*?</td>\s*</tr>\s*</table>",
|
||||
re.DOTALL | re.IGNORECASE
|
||||
)
|
||||
|
||||
with open('README.md', 'r') as f:
|
||||
txt = f.read()
|
||||
|
||||
cleaned = pattern.sub("", txt)
|
||||
print(cleaned)
|
||||
```
|
||||
@@ -1,10 +1,10 @@
|
||||
--extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/
|
||||
packaging==23.2
|
||||
peft==0.9.0
|
||||
transformers @ git+https://github.com/huggingface/transformers.git@f6261d7d81edd036fc53bfede65fe91f01a661aa
|
||||
transformers @ git+https://github.com/huggingface/transformers.git@73a73b415e36f41481369f6129cb4b62bb127a78
|
||||
tokenizers==0.15.0
|
||||
bitsandbytes>=0.43.0
|
||||
accelerate==0.26.1
|
||||
bitsandbytes==0.43.0
|
||||
accelerate==0.28.0
|
||||
deepspeed==0.13.1
|
||||
pydantic==2.6.3
|
||||
addict
|
||||
@@ -40,7 +40,3 @@ gcsfs
|
||||
# adlfs
|
||||
|
||||
trl @ git+https://github.com/huggingface/trl.git@304e208f778a5442c30cdda500348226cdc97d90
|
||||
fastcore>=1.5.29
|
||||
|
||||
lpmm @ git+https://github.com/thu-ml/low-bit-optimizers.git@main
|
||||
yacs
|
||||
|
||||
@@ -1,55 +0,0 @@
|
||||
"""module for building the auto wrap policy for FSDP"""
|
||||
import functools
|
||||
|
||||
from peft import PrefixEncoder, PromptEmbedding, PromptEncoder
|
||||
from torch.distributed.fsdp.wrap import (
|
||||
_or_policy,
|
||||
lambda_auto_wrap_policy,
|
||||
transformer_auto_wrap_policy,
|
||||
)
|
||||
from transformers.models.llama.modeling_llama import LlamaDecoderLayer
|
||||
from transformers.models.mistral.modeling_mistral import MistralDecoderLayer
|
||||
from transformers.models.mixtral.modeling_mixtral import MixtralDecoderLayer
|
||||
|
||||
SUPPORTED_AUTO_WRAP_MODEL_TYPES = [
|
||||
"llama",
|
||||
"mistral",
|
||||
"mixtral",
|
||||
]
|
||||
|
||||
|
||||
def get_wrapping_policy_factory(model_type):
|
||||
if model_type == "llama":
|
||||
layer_to_wrap = LlamaDecoderLayer
|
||||
elif model_type == "mistral":
|
||||
layer_to_wrap = MistralDecoderLayer
|
||||
elif model_type == "mixtral":
|
||||
layer_to_wrap = MixtralDecoderLayer
|
||||
|
||||
def get_wrapping_policy():
|
||||
"""This checks for lora layers (has weight and requires_grad)"""
|
||||
|
||||
def lambda_policy_fn(module):
|
||||
return (
|
||||
len(list(module.named_children())) == 0
|
||||
and getattr(module, "weight", None) is not None
|
||||
and module.weight.requires_grad
|
||||
)
|
||||
|
||||
lambda_policy = functools.partial(
|
||||
lambda_auto_wrap_policy, lambda_fn=lambda_policy_fn
|
||||
)
|
||||
transformer_layer_name = layer_to_wrap
|
||||
transformer_wrap_policy = functools.partial(
|
||||
transformer_auto_wrap_policy,
|
||||
transformer_layer_cls=(
|
||||
PrefixEncoder,
|
||||
PromptEncoder,
|
||||
PromptEmbedding,
|
||||
transformer_layer_name,
|
||||
),
|
||||
)
|
||||
policies = [lambda_policy, transformer_wrap_policy]
|
||||
return functools.partial(_or_policy, policies=policies)
|
||||
|
||||
return get_wrapping_policy
|
||||
@@ -8,28 +8,21 @@ import importlib
|
||||
import importlib.util
|
||||
import logging
|
||||
import math
|
||||
import os
|
||||
import sys
|
||||
from abc import abstractmethod
|
||||
from collections import defaultdict
|
||||
from dataclasses import dataclass, field
|
||||
from functools import wraps
|
||||
from pathlib import Path
|
||||
from typing import Any, Dict, List, Literal, Optional, Tuple, Type, Union
|
||||
from typing import Dict, List, Literal, Optional, Type, Union
|
||||
|
||||
import lpmm
|
||||
import torch
|
||||
import transformers
|
||||
from accelerate import FullyShardedDataParallelPlugin
|
||||
from accelerate.utils import str_to_bool
|
||||
from datasets import Dataset
|
||||
from torch import nn
|
||||
from torch.distributed.fsdp import MixedPrecision
|
||||
from torch.optim.lr_scheduler import OneCycleLR
|
||||
from torch.utils.data import BatchSampler, DataLoader, RandomSampler, SequentialSampler
|
||||
from transformers import (
|
||||
EarlyStoppingCallback,
|
||||
PreTrainedModel,
|
||||
Trainer,
|
||||
TrainerCallback,
|
||||
TrainingArguments,
|
||||
@@ -37,9 +30,8 @@ from transformers import (
|
||||
from transformers.trainer_utils import seed_worker
|
||||
from transformers.utils import is_sagemaker_mp_enabled
|
||||
from trl import DPOTrainer
|
||||
from trl.trainer.utils import pad_to_length
|
||||
|
||||
from axolotl.core.policies.auto_wrap import get_wrapping_policy_factory
|
||||
from axolotl.core.trainers import OptimizerNames
|
||||
from axolotl.loraplus import create_loraplus_optimizer
|
||||
from axolotl.monkeypatch.multipack import SUPPORTED_MULTIPACK_MODEL_TYPES
|
||||
from axolotl.monkeypatch.relora import ReLoRACallback, ReLoRAScheduler
|
||||
@@ -66,9 +58,6 @@ from axolotl.utils.schedulers import (
|
||||
get_cosine_schedule_with_warmup_decay_constant,
|
||||
)
|
||||
|
||||
# monkeypatch so it accepts our custom optimizers
|
||||
transformers.training_args.OptimizerNames = OptimizerNames
|
||||
|
||||
if is_sagemaker_mp_enabled():
|
||||
import smdistributed.modelparallel.torch as smp
|
||||
|
||||
@@ -238,104 +227,26 @@ class AxolotlTrainer(Trainer):
|
||||
if self.args.orpo_alpha:
|
||||
self.loss_fct = torch.nn.CrossEntropyLoss(reduction="none")
|
||||
|
||||
@staticmethod
|
||||
def get_optimizer_cls_and_kwargs(
|
||||
args: TrainingArguments, model: Optional[PreTrainedModel] = None
|
||||
) -> Tuple[Any, Any]:
|
||||
optim_args = {}
|
||||
if args.optim_args:
|
||||
for mapping in args.optim_args.replace(" ", "").split(","):
|
||||
key, value = mapping.split("=")
|
||||
optim_args[key] = value
|
||||
|
||||
optimizer_kwargs = {"lr": args.learning_rate}
|
||||
|
||||
adam_kwargs = {
|
||||
"betas": (args.adam_beta1, args.adam_beta2),
|
||||
"eps": args.adam_epsilon,
|
||||
}
|
||||
|
||||
if args.optim in [
|
||||
OptimizerNames.LPMM_ADAMW_4BIT,
|
||||
OptimizerNames.LPMM_ADAMW_4BIT_FUSED,
|
||||
]:
|
||||
optimizer_cls = lpmm.optim.AdamW
|
||||
optimizer_kwargs.update(adam_kwargs)
|
||||
if args.optim == OptimizerNames.LPMM_ADAMW_4BIT_FUSED:
|
||||
optimizer_kwargs.update({"fused": True})
|
||||
return optimizer_cls, optimizer_kwargs
|
||||
|
||||
return Trainer.get_optimizer_cls_and_kwargs(
|
||||
args,
|
||||
model=model,
|
||||
)
|
||||
|
||||
def create_optimizer(self):
|
||||
if self.args.loraplus_lr_ratio is None:
|
||||
return super().create_optimizer()
|
||||
|
||||
opt_model = self.model_wrapped if is_sagemaker_mp_enabled() else self.model
|
||||
|
||||
if self.optimizer is None: # pylint: disable=access-member-before-definition
|
||||
decay_parameters = self.get_decay_parameter_names(opt_model)
|
||||
optimizer_grouped_parameters = [
|
||||
{
|
||||
"params": [
|
||||
p
|
||||
for n, p in opt_model.named_parameters()
|
||||
if (n in decay_parameters and p.requires_grad)
|
||||
],
|
||||
"weight_decay": self.args.weight_decay,
|
||||
},
|
||||
{
|
||||
"params": [
|
||||
p
|
||||
for n, p in opt_model.named_parameters()
|
||||
if (n not in decay_parameters and p.requires_grad)
|
||||
],
|
||||
"weight_decay": 0.0,
|
||||
},
|
||||
]
|
||||
optimizer_cls, optimizer_kwargs = Trainer.get_optimizer_cls_and_kwargs(
|
||||
self.args,
|
||||
opt_model,
|
||||
)
|
||||
|
||||
(
|
||||
loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
|
||||
loraplus_lr_embedding = getattr(self.args, "loraplus_lr_embedding", None)
|
||||
self.optimizer = create_loraplus_optimizer( # pylint: disable=attribute-defined-outside-init
|
||||
opt_model,
|
||||
optimizer_cls,
|
||||
optimizer_kwargs,
|
||||
) = AxolotlTrainer.get_optimizer_cls_and_kwargs(self.args)
|
||||
|
||||
if self.args.loraplus_lr_ratio:
|
||||
loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
|
||||
loraplus_lr_embedding = getattr(
|
||||
self.args, "loraplus_lr_embedding", None
|
||||
)
|
||||
self.optimizer = create_loraplus_optimizer( # pylint: disable=attribute-defined-outside-init
|
||||
opt_model,
|
||||
optimizer_cls,
|
||||
optimizer_kwargs,
|
||||
loraplus_lr_ratio,
|
||||
loraplus_lr_embedding,
|
||||
)
|
||||
|
||||
else:
|
||||
self.optimizer = ( # pylint: disable=attribute-defined-outside-init
|
||||
optimizer_cls(optimizer_grouped_parameters, **optimizer_kwargs)
|
||||
)
|
||||
|
||||
if optimizer_cls.__name__ == "Adam8bit":
|
||||
import bitsandbytes
|
||||
|
||||
manager = bitsandbytes.optim.GlobalOptimManager.get_instance()
|
||||
|
||||
skipped = 0
|
||||
for module in opt_model.modules():
|
||||
if isinstance(module, nn.Embedding):
|
||||
skipped += sum(
|
||||
{
|
||||
p.data_ptr(): p.numel() for p in module.parameters()
|
||||
}.values()
|
||||
)
|
||||
LOG.info(f"skipped {module}: {skipped/2**20}M params")
|
||||
manager.register_module_override(
|
||||
module, "weight", {"optim_bits": 32}
|
||||
)
|
||||
LOG.debug(f"bitsandbytes: will optimize {module} in fp32")
|
||||
LOG.info(f"skipped: {skipped/2**20}M params")
|
||||
loraplus_lr_ratio,
|
||||
loraplus_lr_embedding,
|
||||
)
|
||||
|
||||
if is_sagemaker_mp_enabled():
|
||||
self.optimizer = smp.DistributedOptimizer( # pylint: disable=attribute-defined-outside-init
|
||||
@@ -562,6 +473,58 @@ class AxolotlTrainer(Trainer):
|
||||
return self.orpo_compute_loss(model, inputs, return_outputs=return_outputs)
|
||||
return super().compute_loss(model, inputs, return_outputs=return_outputs)
|
||||
|
||||
@staticmethod
|
||||
def orpo_concatenate_inputs(inputs, label_pad_token=-100, pad_token=0, device=None):
|
||||
concatenated_batch = {}
|
||||
|
||||
max_length = max(
|
||||
inputs["input_ids"].shape[1], inputs["rejected_input_ids"].shape[1]
|
||||
)
|
||||
# Concatenate positive and negative inputs
|
||||
concatenated_batch["input_ids"] = pad_to_length(
|
||||
inputs["input_ids"], max_length, pad_token
|
||||
)
|
||||
concatenated_batch["rejected_input_ids"] = pad_to_length(
|
||||
inputs["rejected_input_ids"], max_length, pad_token
|
||||
)
|
||||
concatenated_batch["labels"] = pad_to_length(
|
||||
inputs["labels"], max_length, label_pad_token
|
||||
)
|
||||
concatenated_batch["rejected_labels"] = pad_to_length(
|
||||
inputs["rejected_labels"], max_length, label_pad_token
|
||||
)
|
||||
concatenated_batch["attention_mask"] = pad_to_length(
|
||||
inputs["attention_mask"], max_length, 0
|
||||
)
|
||||
concatenated_batch["rejected_attention_mask"] = pad_to_length(
|
||||
inputs["rejected_attention_mask"], max_length, 0
|
||||
)
|
||||
concatenated_batch["prompt_attention_mask"] = pad_to_length(
|
||||
inputs["prompt_attention_mask"], max_length, 0
|
||||
).to(device=device)
|
||||
|
||||
input_ids = torch.cat(
|
||||
[concatenated_batch["input_ids"], concatenated_batch["rejected_input_ids"]],
|
||||
dim=0,
|
||||
).to(device=device)
|
||||
attention_mask = torch.cat(
|
||||
[
|
||||
concatenated_batch["attention_mask"],
|
||||
concatenated_batch["rejected_attention_mask"],
|
||||
],
|
||||
dim=0,
|
||||
).to(device=device)
|
||||
labels = torch.cat(
|
||||
[concatenated_batch["labels"], concatenated_batch["rejected_labels"]], dim=0
|
||||
).to(device=device)
|
||||
|
||||
return {
|
||||
"input_ids": input_ids,
|
||||
"labels": labels,
|
||||
"attention_mask": attention_mask,
|
||||
"prompt_attention_mask": concatenated_batch["prompt_attention_mask"],
|
||||
}
|
||||
|
||||
def orpo_compute_custom_loss(self, logits, labels):
|
||||
logits = logits.contiguous()
|
||||
loss = 0.0
|
||||
@@ -602,45 +565,46 @@ class AxolotlTrainer(Trainer):
|
||||
dim=2,
|
||||
index=(mask * chosen_inputs[:, 1:]).unsqueeze(2),
|
||||
).squeeze(2)
|
||||
return torch.mul(per_token_logps, mask.to(dtype=torch.bfloat16)).sum(dim=1).to(
|
||||
dtype=torch.float64
|
||||
) / mask.sum(dim=1).to(dtype=torch.float64)
|
||||
return torch.mul(per_token_logps, mask).sum(dim=1) / mask.sum(dim=1)
|
||||
|
||||
def orpo_compute_loss(self, model, inputs, return_outputs=False):
|
||||
outputs_neg = model(
|
||||
concat_inputs = AxolotlTrainer.orpo_concatenate_inputs(
|
||||
inputs,
|
||||
label_pad_token=-100,
|
||||
pad_token=self.tokenizer.pad_token_id,
|
||||
device=self.accelerator.device,
|
||||
)
|
||||
|
||||
# Perform a single forward pass
|
||||
outputs = model(
|
||||
**{
|
||||
"input_ids": inputs["rejected_input_ids"],
|
||||
"attention_mask": inputs["rejected_attention_mask"],
|
||||
"labels": inputs["rejected_labels"],
|
||||
},
|
||||
output_hidden_states=True,
|
||||
)
|
||||
outputs_pos = model(
|
||||
**{
|
||||
"input_ids": inputs["input_ids"],
|
||||
"attention_mask": inputs["attention_mask"],
|
||||
"labels": inputs["labels"],
|
||||
"input_ids": concat_inputs["input_ids"],
|
||||
"attention_mask": concat_inputs["attention_mask"],
|
||||
"labels": concat_inputs["labels"],
|
||||
},
|
||||
output_hidden_states=True,
|
||||
)
|
||||
|
||||
# Split the outputs for positive and negative examples
|
||||
outputs_pos, outputs_neg = outputs.logits.chunk(2)
|
||||
|
||||
# Calculate NLL loss
|
||||
pos_loss = self.orpo_compute_custom_loss(
|
||||
logits=outputs_pos.logits, labels=inputs["input_ids"]
|
||||
logits=outputs_pos, labels=concat_inputs["input_ids"].chunk(2)[0]
|
||||
)
|
||||
|
||||
# Calculate Log Probability
|
||||
pos_prob = self.orpo_compute_logps(
|
||||
prompt_attention_mask=inputs["prompt_attention_mask"],
|
||||
chosen_inputs=inputs["input_ids"],
|
||||
chosen_attention_mask=inputs["attention_mask"],
|
||||
logits=outputs_pos.logits,
|
||||
prompt_attention_mask=concat_inputs["prompt_attention_mask"],
|
||||
chosen_inputs=concat_inputs["input_ids"].chunk(2)[0],
|
||||
chosen_attention_mask=concat_inputs["attention_mask"].chunk(2)[0],
|
||||
logits=outputs_pos,
|
||||
)
|
||||
neg_prob = self.orpo_compute_logps(
|
||||
prompt_attention_mask=inputs["prompt_attention_mask"],
|
||||
chosen_inputs=inputs["rejected_input_ids"],
|
||||
chosen_attention_mask=inputs["rejected_attention_mask"],
|
||||
logits=outputs_neg.logits,
|
||||
prompt_attention_mask=concat_inputs["prompt_attention_mask"],
|
||||
chosen_inputs=concat_inputs["input_ids"].chunk(2)[1],
|
||||
chosen_attention_mask=concat_inputs["attention_mask"].chunk(2)[1],
|
||||
logits=outputs_neg,
|
||||
)
|
||||
|
||||
# Calculate log odds
|
||||
@@ -676,51 +640,14 @@ class AxolotlTrainer(Trainer):
|
||||
|
||||
@wraps(Trainer.create_accelerator_and_postprocess)
|
||||
def create_accelerator_and_postprocess(self):
|
||||
rank = int(os.environ.get("LOCAL_RANK", 0))
|
||||
res = super().create_accelerator_and_postprocess()
|
||||
|
||||
if self.args.qlora is False:
|
||||
return res
|
||||
|
||||
# the rest of this method override is specific to fsdp + qlora (for now)
|
||||
sync_module_states = (
|
||||
str_to_bool(os.environ.get("FSDP_SYNC_MODULE_STATES", "True")) == 1
|
||||
)
|
||||
|
||||
mp_policy = None
|
||||
amp = os.environ["ACCELERATE_MIXED_PRECISION"]
|
||||
if amp == "fp16":
|
||||
mp_policy = MixedPrecision(
|
||||
param_dtype=torch.float32,
|
||||
reduce_dtype=torch.float32,
|
||||
buffer_dtype=torch.float32,
|
||||
)
|
||||
elif amp == "bf16":
|
||||
mp_policy = MixedPrecision(
|
||||
param_dtype=torch.float32,
|
||||
reduce_dtype=torch.float32,
|
||||
buffer_dtype=torch.float32,
|
||||
)
|
||||
|
||||
# If somehow we figure out how we want to parameterize we want to autocast buffers...
|
||||
# mp_policy = MixedPrecision(param_dtype=torch.bfloat16, reduce_dtype=torch.bfloat16, buffer_dtype=torch.float32)
|
||||
# load_param_skip_names = ['inv_freq']
|
||||
|
||||
if self.is_fsdp_enabled:
|
||||
wrapping_policy = get_wrapping_policy_factory(self.args.model_type)
|
||||
fsdp_plugin = FullyShardedDataParallelPlugin(
|
||||
auto_wrap_policy=wrapping_policy(),
|
||||
cpu_offload=False,
|
||||
use_orig_params=False,
|
||||
limit_all_gathers=True,
|
||||
param_init_fn=lambda module: module.to_empty(
|
||||
device=torch.device("cuda"), recurse=False
|
||||
)
|
||||
if (rank != 0 and sync_module_states)
|
||||
else None,
|
||||
mixed_precision_policy=mp_policy,
|
||||
)
|
||||
self.accelerator.state.fsdp_plugin = fsdp_plugin
|
||||
if (
|
||||
"limit_all_gathers" in self.args.fsdp_config
|
||||
and self.args.fsdp_config["limit_all_gathers"]
|
||||
):
|
||||
self.accelerator.state.fsdp_plugin.limit_all_gathers = True
|
||||
|
||||
return res
|
||||
|
||||
@@ -878,6 +805,12 @@ class TrainerBuilderBase(abc.ABC):
|
||||
self.model = model
|
||||
self.tokenizer = tokenizer
|
||||
|
||||
# in case the model supports tagging, add the axolotl tag.
|
||||
# This makes sure the tag is correctly pushed even if a user calls
|
||||
# model.push_to_hub instad of trainer.push_to_hub.
|
||||
if hasattr(model, "add_model_tags"):
|
||||
model.add_model_tags(["axolotl"])
|
||||
|
||||
@property
|
||||
def model_ref(self):
|
||||
return self._model_ref
|
||||
@@ -1368,6 +1301,7 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
|
||||
train_dataset=self.train_dataset,
|
||||
eval_dataset=self.eval_dataset,
|
||||
args=training_args,
|
||||
tokenizer=self.tokenizer,
|
||||
data_collator=self.build_collator(training_args, **data_collator_kwargs),
|
||||
eval_data_collator=self.build_collator(
|
||||
training_args, is_eval=True, **data_collator_kwargs
|
||||
|
||||
@@ -1,40 +0,0 @@
|
||||
"""module for trainer helpers like OptimizerNames"""
|
||||
|
||||
from transformers.utils import ExplicitEnum
|
||||
|
||||
|
||||
class OptimizerNames(ExplicitEnum):
|
||||
"""
|
||||
Stores the acceptable string identifiers for optimizers.
|
||||
"""
|
||||
|
||||
ADAMW_HF = "adamw_hf"
|
||||
ADAMW_TORCH = "adamw_torch"
|
||||
ADAMW_TORCH_FUSED = "adamw_torch_fused"
|
||||
ADAMW_TORCH_XLA = "adamw_torch_xla"
|
||||
ADAMW_TORCH_NPU_FUSED = "adamw_torch_npu_fused"
|
||||
ADAMW_APEX_FUSED = "adamw_apex_fused"
|
||||
ADAFACTOR = "adafactor"
|
||||
ADAMW_ANYPRECISION = "adamw_anyprecision"
|
||||
SGD = "sgd"
|
||||
ADAGRAD = "adagrad"
|
||||
ADAMW_BNB = "adamw_bnb_8bit"
|
||||
ADAMW_8BIT = "adamw_8bit" # just an alias for adamw_bnb_8bit
|
||||
LION_8BIT = "lion_8bit"
|
||||
LION = "lion_32bit"
|
||||
PAGED_ADAMW = "paged_adamw_32bit"
|
||||
PAGED_ADAMW_8BIT = "paged_adamw_8bit"
|
||||
PAGED_LION = "paged_lion_32bit"
|
||||
PAGED_LION_8BIT = "paged_lion_8bit"
|
||||
RMSPROP = "rmsprop"
|
||||
RMSPROP_BNB = "rmsprop_bnb"
|
||||
RMSPROP_8BIT = "rmsprop_bnb_8bit"
|
||||
RMSPROP_32BIT = "rmsprop_bnb_32bit"
|
||||
GALORE_ADAMW = "galore_adamw"
|
||||
GALORE_ADAMW_8BIT = "galore_adamw_8bit"
|
||||
GALORE_ADAFACTOR = "galore_adafactor"
|
||||
GALORE_ADAMW_LAYERWISE = "galore_adamw_layerwise"
|
||||
GALORE_ADAMW_8BIT_LAYERWISE = "galore_adamw_8bit_layerwise"
|
||||
GALORE_ADAFACTOR_LAYERWISE = "galore_adafactor_layerwise"
|
||||
LPMM_ADAMW_4BIT = "lmpp_adamw_4bit"
|
||||
LPMM_ADAMW_4BIT_FUSED = "lmpp_adamw_4bit_fused"
|
||||
|
||||
@@ -5,4 +5,4 @@ from functools import partial
|
||||
|
||||
from ..base import load as load_base
|
||||
|
||||
load = partial(load_base, module="axolotl.prompt_strategies.dpo")
|
||||
load = partial(load_base, module_base="axolotl.prompt_strategies.dpo")
|
||||
|
||||
@@ -36,6 +36,7 @@ def load(
|
||||
chat_template = chat_templates(chat_template)
|
||||
except ValueError:
|
||||
pass
|
||||
tokenizer.chat_template = chat_template
|
||||
|
||||
return ORPOTokenizingStrategy(
|
||||
ORPOPrompter(chat_template, tokenizer),
|
||||
|
||||
@@ -119,6 +119,10 @@ def normalize_config(cfg):
|
||||
model_config = load_model_config(cfg)
|
||||
cfg.model_config_type = model_config.model_type
|
||||
|
||||
cfg.tokenizer_config = (
|
||||
cfg.tokenizer_config or cfg.base_model_config or cfg.base_model
|
||||
)
|
||||
|
||||
# figure out if the model is llama
|
||||
cfg.is_llama_derived_model = (
|
||||
(hasattr(model_config, "model_type") and model_config.model_type == "llama")
|
||||
|
||||
@@ -134,7 +134,7 @@ def load_tokenized_prepared_datasets(
|
||||
split="train",
|
||||
) -> Tuple[DatasetDict, List[Prompter]]:
|
||||
cfg_datasets = cfg.test_datasets if split == "test" else cfg.datasets
|
||||
tokenizer_name = tokenizer.__class__.__name__
|
||||
tokenizer_name = cfg.tokenizer_config
|
||||
ds_hash = str(
|
||||
md5(
|
||||
(
|
||||
|
||||
@@ -5,16 +5,14 @@ import logging
|
||||
import math
|
||||
import os
|
||||
import types
|
||||
from typing import Any, Dict, List, Optional, Tuple, Type, Union # noqa: F401
|
||||
from typing import Any, Dict, Optional, Tuple, Union # noqa: F401
|
||||
|
||||
import addict
|
||||
import bitsandbytes as bnb
|
||||
import safetensors
|
||||
import torch
|
||||
import transformers
|
||||
from accelerate import init_empty_weights
|
||||
from bitsandbytes.nn import Linear4bit, Params4bit
|
||||
from fastcore.parallel import parallel
|
||||
from bitsandbytes.nn import Params4bit
|
||||
from peft import (
|
||||
LoftQConfig,
|
||||
PeftConfig,
|
||||
@@ -23,7 +21,7 @@ from peft import (
|
||||
prepare_model_for_kbit_training,
|
||||
)
|
||||
from peft.tuners.lora import QuantLinear
|
||||
from torch import Tensor, nn
|
||||
from torch import nn
|
||||
from transformers import ( # noqa: F401
|
||||
AddedToken,
|
||||
AutoConfig,
|
||||
@@ -35,9 +33,7 @@ from transformers import ( # noqa: F401
|
||||
PreTrainedTokenizerBase,
|
||||
)
|
||||
from transformers.integrations.deepspeed import is_deepspeed_zero3_enabled
|
||||
from transformers.utils import SAFE_WEIGHTS_INDEX_NAME, SAFE_WEIGHTS_NAME, hub
|
||||
|
||||
from axolotl.core.policies.auto_wrap import SUPPORTED_AUTO_WRAP_MODEL_TYPES
|
||||
from axolotl.models.mamba import fix_mamba_attn_for_loss
|
||||
from axolotl.monkeypatch.multipack import (
|
||||
SUPPORTED_MULTIPACK_MODEL_TYPES,
|
||||
@@ -138,9 +134,8 @@ def load_tokenizer(cfg):
|
||||
if cfg.tokenizer_type:
|
||||
tokenizer_cls = getattr(transformers, cfg.tokenizer_type)
|
||||
|
||||
tokenizer_config = cfg.tokenizer_config or cfg.base_model_config or cfg.base_model
|
||||
tokenizer = tokenizer_cls.from_pretrained(
|
||||
tokenizer_config,
|
||||
cfg.tokenizer_config,
|
||||
trust_remote_code=cfg.trust_remote_code or False,
|
||||
use_fast=use_fast,
|
||||
**tokenizer_kwargs,
|
||||
@@ -272,117 +267,6 @@ def load_tokenizer(cfg):
|
||||
return tokenizer
|
||||
|
||||
|
||||
def replace_linear(
|
||||
model: nn.Module,
|
||||
linear_replacement: Type[nn.Module],
|
||||
quant_config: Union[dict, None] = None,
|
||||
skip_modules=None,
|
||||
**kwargs,
|
||||
):
|
||||
"""
|
||||
Replace linear modules with a new Linear module.
|
||||
Parameters:
|
||||
model (`torch.nn.Module`):
|
||||
Input model or `torch.nn.Module` as the function is run recursively.
|
||||
linear_replacement (`torch.nn.Module`):
|
||||
The linear module that replaces the old one. Only expects standard arguments.
|
||||
If other arguments need to be passed, use a lambda.
|
||||
skip_modules (`List[str]`, *optional*, defaults to `lm_head`):
|
||||
List of modules names not to convert. Defaults to `lm_head`.
|
||||
"""
|
||||
if skip_modules is None:
|
||||
skip_modules = ["lm_head"]
|
||||
for name, module in model.named_children():
|
||||
if len(list(module.children())) > 0:
|
||||
replace_linear(
|
||||
module, linear_replacement, quant_config, skip_modules, **kwargs
|
||||
)
|
||||
|
||||
if isinstance(module, torch.nn.Linear) and name not in skip_modules:
|
||||
if issubclass(linear_replacement, Linear4bit):
|
||||
model._modules[ # pylint: disable=protected-access
|
||||
name
|
||||
] = linear_replacement(
|
||||
module.in_features,
|
||||
module.out_features,
|
||||
module.bias is not None,
|
||||
**kwargs,
|
||||
)
|
||||
else:
|
||||
raise ValueError(
|
||||
f"Unsupported linear replacement: {type(linear_replacement)}"
|
||||
)
|
||||
return model
|
||||
|
||||
|
||||
def load_and_quantize(
|
||||
module: nn.Module,
|
||||
name: str,
|
||||
value: Tensor,
|
||||
device: torch.device = None,
|
||||
dtype: torch.dtype = None,
|
||||
skip_names: Optional[List[str]] = None,
|
||||
is_meta_rank: bool = False,
|
||||
low_memory: bool = True,
|
||||
verbose: bool = False,
|
||||
quant_method: str = "bnb",
|
||||
):
|
||||
"""
|
||||
Loads `value` tensor into submodule of `module`, optionally skipping `skip_names` and converting to `dtype`.
|
||||
|
||||
Quantizes `Params4bit` on `device` then places on "cpu" if low_memory=True or "meta" if is_meta_rank=True.
|
||||
"""
|
||||
|
||||
if skip_names is None:
|
||||
skip_names = []
|
||||
|
||||
def place_on_device(value):
|
||||
if is_meta_rank:
|
||||
device = "meta"
|
||||
elif low_memory:
|
||||
device = "cpu"
|
||||
else:
|
||||
device = "cuda"
|
||||
return value.to(device=device, dtype=dtype)
|
||||
|
||||
if any(skip_name in name for skip_name in skip_names):
|
||||
if verbose:
|
||||
print(f"Skipping {name} because it is in skip_names")
|
||||
return
|
||||
|
||||
module_key, _, value_key = name.rpartition(".")
|
||||
try:
|
||||
submodule = module.get_submodule(module_key)
|
||||
except AttributeError as exc:
|
||||
print(f"Module {module_key} not found:\n{exc}")
|
||||
return
|
||||
|
||||
try:
|
||||
if quant_method == "bnb":
|
||||
param = submodule.get_parameter(value_key)
|
||||
if isinstance(param, Params4bit):
|
||||
# With `sync_module_states=True`, a meta device Params4bit needs to be the same
|
||||
# shape as the quantized Params4bit with an initialized quant_state. However,
|
||||
# FSDP only syncs parameters and buffers, so the quant_state isn't copied. This
|
||||
# workaround quantizes Params4bit to initialize quant_state on all ranks, then
|
||||
# replaces Params4bit's data with a meta tensor to free memory on non-rank 0.
|
||||
value = type(param)(
|
||||
value.to(device=device, dtype=dtype).data, **param.__dict__
|
||||
).cuda(device)
|
||||
if is_meta_rank:
|
||||
value = type(param)(value.data.to("meta"), **value.__dict__)
|
||||
elif low_memory:
|
||||
value = type(param)(value.data.to("cpu"), **value.__dict__)
|
||||
else:
|
||||
value = type(param)(place_on_device(value).data)
|
||||
|
||||
except AttributeError:
|
||||
# it's a buffer
|
||||
value = place_on_device(value)
|
||||
|
||||
setattr(submodule, value_key, value)
|
||||
|
||||
|
||||
def load_model(
|
||||
cfg: DictDefault,
|
||||
tokenizer: PreTrainedTokenizerBase,
|
||||
@@ -568,6 +452,7 @@ def load_model(
|
||||
"bnb_4bit_compute_dtype": cfg.torch_dtype,
|
||||
"bnb_4bit_use_double_quant": True,
|
||||
"bnb_4bit_quant_type": "nf4",
|
||||
"bnb_4bit_quant_storage": torch.bfloat16,
|
||||
}
|
||||
|
||||
if cfg.bnb_config_kwargs:
|
||||
@@ -617,78 +502,10 @@ def load_model(
|
||||
model_kwargs["attn_implementation"] = "eager"
|
||||
model_config._attn_implementation = "eager" # pylint: disable=protected-access
|
||||
|
||||
qlora_fsdp = (
|
||||
cfg.fsdp
|
||||
and cfg.adapter == "qlora"
|
||||
and model_config.model_type in SUPPORTED_AUTO_WRAP_MODEL_TYPES
|
||||
)
|
||||
qlora_fsdp = cfg.fsdp and cfg.adapter == "qlora"
|
||||
|
||||
try:
|
||||
if qlora_fsdp:
|
||||
if cfg.bf16 or cfg.bfloat16:
|
||||
torch_dtype, compute_dtype = torch.float32, torch.bfloat16
|
||||
elif cfg.fp16 or cfg.float16:
|
||||
torch_dtype, compute_dtype = torch.float32, torch.float16
|
||||
else:
|
||||
torch_dtype, compute_dtype = torch.float32, torch.float16
|
||||
|
||||
with init_empty_weights():
|
||||
LOG.info("Loading model with empty weights.")
|
||||
model = AutoModelForCausalLM.from_config(model_config)
|
||||
model.model = replace_linear(
|
||||
model.model,
|
||||
Linear4bit,
|
||||
compute_dtype=compute_dtype,
|
||||
quant_type="nf4",
|
||||
quant_storage=torch_dtype,
|
||||
)
|
||||
|
||||
model.is_loaded_in_4bit = True
|
||||
|
||||
# Grab the safetensors files that hold the weights
|
||||
try:
|
||||
idx = hub.cached_file(base_model, SAFE_WEIGHTS_INDEX_NAME)
|
||||
files, _ = hub.get_checkpoint_shard_files(base_model, idx)
|
||||
except OSError:
|
||||
try:
|
||||
# This means the model doesn't have a model.safetensors.index.json because it is not sharded
|
||||
files = []
|
||||
files.append(hub.cached_file(base_model, SAFE_WEIGHTS_NAME))
|
||||
except OSError as exc:
|
||||
# This means the model probably doesn't have a safetensors file
|
||||
raise exc
|
||||
|
||||
# Load in the weights, using our custom load_and_quantize method which quantizes Params4bit on the fly
|
||||
# and then places each layer on CPU or meta if using low_memory to minimize GPU memory usage
|
||||
def load_and_quantize_parallel(name_param, model, **kwargs):
|
||||
name, param = name_param
|
||||
load_and_quantize(model, name, param, **kwargs)
|
||||
|
||||
param_count = sum((p.numel() for n, p in model.named_parameters()))
|
||||
for filename in files:
|
||||
weights = safetensors.torch.load_file(filename)
|
||||
quant_method = "bnb"
|
||||
devprops = torch.cuda.get_device_properties(torch.cuda.current_device())
|
||||
left = int(os.cpu_count() / torch.cuda.device_count())
|
||||
right = int(
|
||||
8 * (devprops.total_memory / 1e9 / 40) * (70 / (param_count / 1e9))
|
||||
)
|
||||
n_workers = min(left, right)
|
||||
parallel(
|
||||
load_and_quantize_parallel,
|
||||
weights.items(),
|
||||
n_workers=n_workers,
|
||||
threadpool=True,
|
||||
model=model,
|
||||
dtype=torch_dtype,
|
||||
device=cfg.local_rank,
|
||||
skip_names=[],
|
||||
is_meta_rank=(cfg.local_rank != 0),
|
||||
verbose=False,
|
||||
quant_method=quant_method,
|
||||
)
|
||||
|
||||
elif (
|
||||
if (
|
||||
model_config.model_type == "llama"
|
||||
and not cfg.trust_remote_code
|
||||
and not cfg.gptq
|
||||
@@ -715,32 +532,6 @@ def load_model(
|
||||
if cfg.flash_attn_fuse_qkv:
|
||||
LOG.info("patching with fused QKV")
|
||||
replace_llama_qkv_with_fused(model)
|
||||
# elif model_type == "GPTNeoXForCausalLM" and cfg.flash_attention:
|
||||
# This is a WIP, still an issue with the backward pass
|
||||
# RuntimeError: grad can be implicitly created only for scalar outputs
|
||||
# TODO: try config.sequence_parallel = False
|
||||
# # https://github.com/HazyResearch/flash-attention/blob/40a25c8ee7465cf547b929cfa2937034e37bfce9/tests/models/test_gpt_neox.py#L12
|
||||
# # https://github.com/HazyResearch/flash-attention/tree/main/training#model-components
|
||||
# # add `**kwargs` to https://github.com/HazyResearch/flash-attention/blob/40a25c8ee7465cf547b929cfa2937034e37bfce9/flash_attn/models/gpt.py#L442
|
||||
# from flash_attn.utils.pretrained import state_dict_from_pretrained
|
||||
# from flash_attn.models.gpt import GPTLMHeadModel
|
||||
# from flash_attn.models.gpt_neox import remap_state_dict_hf_gpt_neox, gpt_neox_config_to_gpt2_config
|
||||
# from transformers import GPTNeoXConfig
|
||||
# config = gpt_neox_config_to_gpt2_config(GPTNeoXConfig.from_pretrained(base_model))
|
||||
# config.use_flash_attn = True
|
||||
# config.fused_bias_fc = True
|
||||
# config.fused_mlp = True # GPT-NeoX-20B uses "gelu_fast"
|
||||
# config.activation_function = "gelu_fast"
|
||||
# config.fused_dropout_add_ln = True
|
||||
# # config.residual_in_fp32 = True
|
||||
#
|
||||
# model: GPTLMHeadModel = GPTLMHeadModel.from_pretrained(
|
||||
# base_model,
|
||||
# config,
|
||||
# dtype=torch_dtype,
|
||||
# device=cfg.device,
|
||||
# )
|
||||
# model.train() # sets to train instead of eval mode
|
||||
elif model_type == "MambaLMHeadModel":
|
||||
# FIXME this is janky at best and hacked together to make it work
|
||||
MambaLMHeadModel = fix_mamba_attn_for_loss() # pylint: disable=invalid-name
|
||||
|
||||
@@ -304,6 +304,10 @@ def setup_fsdp_envs(cfg):
|
||||
os.environ["FSDP_OFFLOAD_PARAMS"] = "true"
|
||||
if cfg.fsdp_config.fsdp_sync_module_states:
|
||||
os.environ["FSDP_SYNC_MODULE_STATES"] = "true"
|
||||
if cfg.fsdp_config.fsdp_cpu_ram_efficient_loading:
|
||||
os.environ["FSDP_CPU_RAM_EFFICIENT_LOADING"] = "true"
|
||||
if cfg.fsdp_config.fsdp_use_orig_params:
|
||||
os.environ["FSDP_USE_ORIG_PARAMS"] = "true"
|
||||
if cfg.fsdp_config.fsdp_state_dict_type:
|
||||
os.environ["FSDP_STATE_DICT_TYPE"] = cfg.fsdp_config.fsdp_state_dict_type
|
||||
if cfg.fsdp_config.fsdp_transformer_layer_cls_to_wrap:
|
||||
|
||||
1
styles.css
Normal file
1
styles.css
Normal file
@@ -0,0 +1 @@
|
||||
/* css styles */
|
||||
@@ -1,16 +1,18 @@
|
||||
"""
|
||||
unit tests for axolotl.core.trainer_builder
|
||||
"""
|
||||
|
||||
import pytest
|
||||
|
||||
from axolotl.core.trainer_builder import HFDPOTrainerBuilder
|
||||
from axolotl.utils.config import normalize_config
|
||||
from axolotl.utils.dict import DictDefault
|
||||
from axolotl.utils.models import load_model, load_tokenizer
|
||||
|
||||
|
||||
@pytest.fixture(name="cfg")
|
||||
def fixture_cfg():
|
||||
return DictDefault(
|
||||
cfg = DictDefault(
|
||||
{
|
||||
"base_model": "TinyLlama/TinyLlama-1.1B-Chat-v0.6",
|
||||
"model_type": "AutoModelForCausalLM",
|
||||
@@ -34,6 +36,10 @@ def fixture_cfg():
|
||||
}
|
||||
)
|
||||
|
||||
normalize_config(cfg)
|
||||
|
||||
return cfg
|
||||
|
||||
|
||||
@pytest.fixture(name="tokenizer")
|
||||
def fixture_tokenizer(cfg):
|
||||
|
||||
@@ -77,7 +77,7 @@ class TestMixtral(unittest.TestCase):
|
||||
model, _ = train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
|
||||
assert (
|
||||
model.base_model.model.model.layers[0].block_sparse_moe.gate.weight.dtype
|
||||
== torch.uint8
|
||||
== torch.float32
|
||||
)
|
||||
assert (Path(temp_dir) / "adapter_model.bin").exists()
|
||||
|
||||
@@ -131,7 +131,7 @@ class TestMixtral(unittest.TestCase):
|
||||
model, _ = train(cfg=cfg, cli_args=cli_args, dataset_meta=dataset_meta)
|
||||
assert (
|
||||
model.base_model.model.model.layers[0].block_sparse_moe.gate.weight.dtype
|
||||
== torch.uint8
|
||||
== torch.float32
|
||||
)
|
||||
assert (Path(temp_dir) / "adapter_model.bin").exists()
|
||||
|
||||
|
||||
Reference in New Issue
Block a user