Dan Saunders 3d8425fa91 Activation function Triton kernels, LoRA custom autograd functions (#2324) ...

* LoRA + activation fn Triton kernels: initial commit

* implementing optims

* finalizing MLP LoRA kernels and progress on QKV / W kernels

* updates

* O projection optim

* adding monkey patching logic

* doc strings, typing, pre-commit fixes

* updates

* adding lora 8b kernels example

* working on fsdp support

* tests and fixes

* small fixes, getting tests to pass, adding doc strings

* integration tests for LoRA patching

* config.qmd

* remove unneeded pytest fixture

* fix

* review comments first pass

* improving tests, attention class agnostic patching

* adding support for more archs

* wip SiLU / GELU impls

* improved testing, small updates, etc.

* slightly updating docs

* rebase

* fixing test_attention_patching_integration

* additional review comments, fixing test in CI (hopefully)

* isolating problematic patching test

* relaxing allclose threshold to reduce flakiness

* fixing accidental change

* adding model arch agnostic attention class fetching

* removing unused activations

2025-02-17 14:23:15 -05:00

..

cli

[feature] sweeps (#2171)

2025-02-01 21:11:18 -05:00

core

updating to fused (#2293)

2025-01-30 11:45:56 -05:00

e2e

Activation function Triton kernels, LoRA custom autograd functions (#2324)

2025-02-17 14:23:15 -05:00

fixtures

Respect sequence_len in config for type: llama2_chat (#926)

2023-12-12 09:39:22 -08:00

integrations

rename liger test so it properly runs in ci (#2246)

2025-01-09 17:31:43 -05:00

monkeypatch

support for true batches with multipack (#1230)

2024-02-01 10:18:42 -05:00

patched

support for latest transformers release 4.48.1 (#2256)

2025-01-23 21:17:57 -05:00

prompt_strategies

Process reward models (#2241)

2025-01-29 00:08:33 -05:00

utils

Refactor func load_model to class ModelLoader (#1909)

2024-10-25 09:06:56 -04:00

conftest.py

update upstream HF deps (#2239)

2025-01-09 21:01:59 +00:00

constants.py

Add Exact Deduplication Feature to Preprocessing Pipeline (#2072)

2024-12-02 08:47:10 -05:00

test_data.py

CLI Implementation with Click (#2107)

2024-12-05 22:11:48 -05:00

test_datasets.py

rename references to dpo dataset prep to pref data (#2258)

2025-01-14 22:07:55 -05:00

test_dict.py

Pydantic 2.x cfg (#1239)

2024-02-26 12:24:14 -05:00

test_exact_deduplication.py

rename references to dpo dataset prep to pref data (#2258)

2025-01-14 22:07:55 -05:00

test_expand_mask.py

Attention mask and position id fixes for packing (#285)

2023-08-12 15:14:56 -04:00

test_freeze.py

Train parameters exclusively in specific ranges (#1390)

2024-03-14 11:05:42 -04:00

test_lora.py

assume empty lora dropout means 0.0 and add tests (#2243)

2025-01-13 10:44:11 -05:00

test_normalize_config.py

remove fastchat and sharegpt (#2021)

2024-11-08 13:45:49 -05:00

test_packed_batch_sampler.py

Switch to parallel FFD bin packing algorithm. (#1619)

2024-05-23 17:32:14 -04:00

test_packed_dataset.py

Attention mask and position id fixes for packing (#285)

2023-08-12 15:14:56 -04:00

test_packed_pretraining.py

Pretrain multipack (#2278)

2025-01-24 12:55:20 -05:00

test_perplexity.py

various tests fixes for flakey tests (#2110)

2024-12-02 17:28:58 -05:00

test_prompt_tokenizers.py

rename liger test so it properly runs in ci (#2246)

2025-01-09 17:31:43 -05:00

test_prompters.py

fix: prompt phi (#1845) [skip ci]

2024-08-22 11:46:57 -04:00

test_schedulers.py

add optimizer step to prevent warning in tests (#1502) [skip ci]

2024-11-19 10:19:03 -05:00

test_tokenizers.py

Support for additional_special_tokens (#1221) [skip ci]

2024-01-31 18:13:13 -05:00

test_validation_dataset.py

Check torch version for ADOPT optimizer + integrating new ADOPT updates (#2104)

2024-12-02 20:15:39 -05:00