pre-commit formatting fixes

Command had accidentally been moved out of if-else block.

.github/workflows/base.yml

@@ -12,28 +12,19 @@ jobs:
     # this job needs to be run on self-hosted GPU runners...
     runs-on: self-hosted
     strategy:
+      fail-fast: false
       matrix:
         include:
           - cuda: "118"
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
-            axolotl_extras:
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
           - cuda: "118"
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
-            axolotl_extras:
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
-          - cuda: "117"
-            cuda_version: 11.7.0
-            python_version: "3.9"
-            pytorch: 1.13.1
-            axolotl_extras:
-          - cuda: "118"
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.0
-            axolotl_extras: gptq
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -57,11 +48,9 @@ jobs:
           push: ${{ github.event_name != 'pull_request' }}
           tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
           build-args: |
             CUDA_VERSION=${{ matrix.cuda_version }}
             CUDA=${{ matrix.cuda }}
             PYTHON_VERSION=${{ matrix.python_version }}
             PYTORCH_VERSION=${{ matrix.pytorch }}
-            AXOLOTL_EXTRAS=${{ matrix.axolotl_extras }}
+            TORCH_CUDA_ARCH_LIST=${{ matrix.torch_cuda_arch_list }}

.github/workflows/main.yml

@@ -11,28 +11,24 @@ jobs:
     if: github.repository_owner == 'OpenAccess-AI-Collective'
     # this job needs to be run on self-hosted GPU runners...
     strategy:
+      fail-fast: false
       matrix:
         include:
           - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras:
           - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras:
           - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras: gptq
-          - cuda: cu117
-            cuda_version: 11.7.0
-            python_version: "3.9"
-            pytorch: 1.13.1
-            axolotl_extras:
     runs-on: self-hosted
     steps:
       - name: Checkout
@@ -54,13 +50,11 @@ jobs:
         with:
           context: .
           build-args: |
-            BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}
           file: ./docker/Dockerfile
           push: ${{ github.event_name != 'pull_request' }}
           tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
   build-axolotl-runpod:
     needs: build-axolotl
     if: github.repository_owner == 'OpenAccess-AI-Collective'
@@ -68,26 +62,21 @@ jobs:
     strategy:
       matrix:
         include:
-          - cuda: cu118
+          - cuda: 118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras:
-          - cuda: cu118
+          - cuda: 118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras:
-          - cuda: cu118
+          - cuda: 118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.0
+            pytorch: 2.0.1
             axolotl_extras: gptq
-          - cuda: cu117
-            cuda_version: 11.7.0
-            python_version: "3.9"
-            pytorch: 1.13.1
-            axolotl_extras:
     runs-on: self-hosted
     steps:
       - name: Checkout
@@ -109,10 +98,9 @@ jobs:
         with:
           context: .
           build-args: |
-            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            CUDA=${{ matrix.cuda }}
           file: ./docker/Dockerfile-runpod
           push: ${{ github.event_name != 'pull_request' }}
-          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
-          cache-from: type=gha
-          cache-to: type=gha,mode=max

.github/workflows/tests.yml

@@ -7,6 +7,7 @@ jobs:
   test:
     runs-on: ubuntu-latest
     strategy:
+      fail-fast: false
       matrix:
         python_version: ["3.9", "3.10"]
     timeout-minutes: 10

.pre-commit-config.yaml

@@ -1,5 +1,5 @@
 default_language_version:
-    python: python3.9
+    python: python3
 
 repos:
 -   repo: https://github.com/pre-commit/pre-commit-hooks

LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -24,11 +24,12 @@
 | mpt      | ✅         | ❌    | ❓     | ❌    | ❓            | ❌                 | ❌          | ❓             |
 | falcon   | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❌          | ✅             |
 | gpt-j    | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❓          | ✅             |
+| XGen     | ✅         | ❓    | ✅     | ❓    | ❓            | ❓                 | ❓          | ✅
 
 
 ## Quickstart ⚡
 
-**Requirements**: Python 3.9 and Pytorch 2.0.
+**Requirements**: Python >=3.9 and Pytorch >=2.0.
 
 ```bash
 git clone https://github.com/OpenAccess-AI-Collective/axolotl
@@ -36,8 +37,6 @@ git clone https://github.com/OpenAccess-AI-Collective/axolotl
 pip3 install -e .
 pip3 install -U git+https://github.com/huggingface/peft.git
 
-accelerate config
-
 # finetune lora
 accelerate launch scripts/finetune.py examples/openllama-3b/lora.yml
 
@@ -52,11 +51,10 @@ accelerate launch scripts/finetune.py examples/openllama-3b/lora.yml \
 
 - Docker
   ```bash
-  docker run --gpus '"all"' --rm -it winglian/axolotl:main-py3.9-cu118-2.0.0
+  docker run --gpus '"all"' --rm -it winglian/axolotl:main-py3.10-cu118-2.0.1
   ```
-  - `winglian/axolotl-runpod:main-py3.9-cu118-2.0.0`: for runpod
+  - `winglian/axolotl-runpod:main-py3.10-cu118-2.0.1`: for runpod
-  - `winglian/axolotl-runpod:main-py3.9-cu118-2.0.0-gptq`: for gptq
+  - `winglian/axolotl-runpod:main-py3.9-cu118-2.0.1-gptq`: for gptq
-  - `winglian/axolotl:dev`: dev branch (not usually up to date)
 
   Or run on the current files for development:
 
@@ -108,7 +106,7 @@ accelerate launch scripts/finetune.py examples/openllama-3b/lora.yml \
 
   3. Install torch
   ```bash
-  pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+  pip3 install -U torch --index-url https://download.pytorch.org/whl/cu118
   ```
 
   4. Axolotl
@@ -138,7 +136,7 @@ Have dataset(s) in one of the following format (JSONL recommended):
   ```json
   {"instruction": "...", "input": "...", "output": "..."}
   ```
-- `sharegpt`: conversations
+- `sharegpt:chat`: conversations
   ```json
   {"conversations": [{"from": "...", "value": "..."}]}
   ```
@@ -195,6 +193,10 @@ Have dataset(s) in one of the following format (JSONL recommended):
   ```json
   {"message_1": "...", "message_2": "..."}
   ```
+- `alpaca_w_system.load_open_orca`: support for open orca datasets with included system prompts, instruct
+  ```json
+  {"system_prompt": "...", "question": "...", "response": "..."}
+  ```
 - `context_qa`: in context question answering from an article
   ```json
   {"article": "...", "question": "...", "answer": "..."}
@@ -233,7 +235,7 @@ Have dataset(s) in one of the following format (JSONL recommended):
 #### How to add custom prompts
 
   1. Add your method to a file in [prompt_strategies](src/axolotl/prompt_strategies). Please see other files as example.
-  2. Use your custom file name as the dataset type.
+  2. Use your custom file name as the dataset type `<prompt_strategies_file>.load_<load_fn>`.
 
 Optionally, download some datasets, see [data/README.md](data/README.md)
 
@@ -241,7 +243,7 @@ Optionally, download some datasets, see [data/README.md](data/README.md)
 
 ### Config
 
-See sample configs in [configs](configs) folder or [examples](examples) for quick start. It is recommended to duplicate and modify to your needs. The most important options are:
+See [examples](examples) for quick start. It is recommended to duplicate and modify to your needs. The most important options are:
 
 - model
   ```yaml
@@ -251,10 +253,24 @@ See sample configs in [configs](configs) folder or [examples](examples) for quic
 
 - dataset
   ```yaml
+  sequence_len: 2048 # max token length for prompt
+
+  # huggingface repo
   datasets:
-    - path: vicgalle/alpaca-gpt4 # local or huggingface repo
+    - path: vicgalle/alpaca-gpt4
+      type: alpaca # format from earlier
+
+  # huggingface repo with specific configuration/subset
+  datasets:
+    - path: EleutherAI/pile
+      name: enron_emails
+      type: completion # format from earlier
+
+  # local
+  datasets:
+    - path: json
+      data_files: data.jsonl # or json
       type: alpaca # format from earlier
-  sequence_len: 2048 # max token length / prompt
   ```
 
 - loading
@@ -264,6 +280,8 @@ See sample configs in [configs](configs) folder or [examples](examples) for quic
   bf16: true # require >=ampere
   fp16: true
   tf32: true # require >=ampere
+  bfloat16: true # require >=ampere, use instead of bf16 when you don't want AMP (automatic mixed precision)
+  float16: true # use instead of fp16 when you don't want AMP
   ```
   Note: Repo does not do 4-bit quantization.
 
@@ -291,6 +309,8 @@ base_model_ignore_patterns:
 # if the base_model repo on hf hub doesn't include configuration .json files,
 # you can set that here, or leave this empty to default to base_model
 base_model_config: ./llama-7b-hf
+# you can specify to choose a specific model revision from huggingface hub
+model_revision:
 # Optional tokenizer configuration override in case you want to use a different tokenizer
 # than the one defined in the base model
 tokenizer_config:
@@ -300,6 +320,11 @@ model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
 # Trust remote code for untrusted source
 trust_remote_code:
+# use_fast option for tokenizer loading from_pretrained, default to True
+tokenizer_use_fast:
+# resize the model embeddings when new tokens are added to multiples of 32
+# this is reported to improve training speed on some models
+resize_token_embeddings_to_32x:
 
 # whether you are training a 4-bit GPTQ quantized model
 gptq: true
@@ -320,18 +345,21 @@ tf32: true # require >=ampere
 
 # a list of one or more datasets to finetune the model with
 datasets:
-  # this can be either a hf dataset, or relative path
+  # hf dataset repo | "json" for local dataset, make sure to fill data_files
   - path: vicgalle/alpaca-gpt4
   # The type of prompt to use for training. [alpaca, sharegpt, gpteacher, oasst, reflection]
-    type: alpaca # format OR format:prompt_style (chat/instruct)
+    type: alpaca # format | format:<prompt_style> (chat/instruct) | <prompt_strategies>.load_<load_fn>
     data_files: # path to source data files
     shards: # number of shards to split data into
+    name: # name of dataset configuration to load
 
 # axolotl attempts to save the dataset as an arrow after packing the data together so
 # subsequent training attempts load faster, relative path
 dataset_prepared_path: data/last_run_prepared
 # push prepared dataset to hub
 push_dataset_to_hub: # repo path
+# push checkpoints to hub
+hub_model_id: # repo path to push finetuned model
 # whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets
 # required to be true when used in combination with `push_dataset_to_hub`
 hf_use_auth_token: # boolean
@@ -393,6 +421,9 @@ logging_steps:
 save_steps:
 eval_steps:
 
+# save model as safetensors (require safetensors package)
+save_safetensors:
+
 # whether to mask out or include the human's prompt from the training labels
 train_on_inputs: false
 # don't use this, leads to wonky training (according to someone on the internet)
@@ -420,7 +451,15 @@ log_sweep_max_lr:
 optimizer:
 # specify weight decay
 weight_decay:
+# adamw hyperparams
+adam_beta1:
+adam_beta2:
+adam_epsilon:
+# Gradient clipping max norm
+max_grad_norm:
 
+# whether to bettertransformers
+flash_optimum:
 # whether to use xformers attention patch https://github.com/facebookresearch/xformers:
 xformers_attention:
 # whether to use flash attention patch https://github.com/HazyResearch/flash-attention:
@@ -476,17 +515,6 @@ strict:
 
 </details>
 
-### Accelerate
-
-Configure accelerate
-
-```bash
-accelerate config
-
-# Edit manually
-# nano ~/.cache/huggingface/accelerate/default_config.yaml
-```
-
 ### Train
 
 Run
@@ -494,6 +522,21 @@ Run
 accelerate launch scripts/finetune.py configs/your_config.yml
 ```
 
+#### Multi-GPU Config
+
+- llama FSDP
+```yaml
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_offload_params: true
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
+```
+
+- llama Deepspeed: append `ACCELERATE_USE_DEEPSPEED=true` in front of finetune command
+
 ### Inference
 
 Pass the appropriate flag to the train command:
@@ -520,6 +563,12 @@ Add below flag to train command above
 --merge_lora --lora_model_dir="./completed-model" --load_in_8bit=False --load_in_4bit=False
 ```
 
+If you run out of CUDA memory, you can try to merge in system RAM with
+
+```bash
+CUDA_VISIBLE_DEVICES="" python3 scripts/finetune.py ...
+```
+
 ## Common Errors 🧰
 
 > Cuda out of memory
@@ -538,6 +587,10 @@ Try set `fp16: true`
 
 Try to turn off xformers.
 
+> accelerate config missing
+
+It's safe to ignore it.
+
 ## Need help? 🙋♂️
 
 Join our [Discord server](https://discord.gg/HhrNrHJPRb) where we can help you

data/README.md

@@ -10,10 +10,10 @@ curl https://github.com/teknium1/GPTeacher/blob/main/Roleplay/roleplay-similarit
 ## Convert the JSON data files to JSONL.
 
 ```shell
-python3 ./scripts/alpaca_json_to_jsonl.py --input data/alpaca_data_gpt4.json > data/alpaca_data_gpt4.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/alpaca_data_gpt4.json --output data/alpaca_data_gpt4.jsonl
-python3 ./scripts/alpaca_json_to_jsonl.py --input data/raw/vicuna_cleaned.json > data/vicuna_cleaned.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/vicuna_cleaned.json --output data/vicuna_cleaned.jsonl
-python3 ./scripts/alpaca_json_to_jsonl.py --input data/raw/roleplay-similarity_0.6-instruct-dataset.json > data/roleplay-similarity_0.6-instruct-dataset.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/roleplay-similarity_0.6-instruct-dataset.json --output data/roleplay-similarity_0.6-instruct-dataset.jsonl
-python3 ./scripts/alpaca_json_to_jsonl.py --input data/raw/gpt4-instruct-similarity-0.6-dataset.json > data/gpt4-instruct-similarity-0.6-dataset.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/gpt4-instruct-similarity-0.6-dataset.json --output data/gpt4-instruct-similarity-0.6-dataset.jsonl
 ```
 ---
 

docker/Dockerfile

@@ -3,16 +3,15 @@ FROM winglian/axolotl-base:$BASE_TAG
 
 ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 ARG AXOLOTL_EXTRAS=""
+ARG CUDA="118"
+ENV BNB_CUDA_VERSION=$CUDA
 
 RUN apt-get update && \
     apt-get install -y vim curl
 
 WORKDIR /workspace
 
-RUN pip3 install --force-reinstall "peft @ git+https://github.com/huggingface/peft.git@main" \
+RUN pip3 install --force-reinstall "peft @ git+https://github.com/huggingface/peft.git@main"
-            "accelerate @ git+https://github.com/huggingface/accelerate.git@main" \
-            "transformers @ git+https://github.com/huggingface/transformers.git@main"
-
 RUN git clone --depth=1 https://github.com/OpenAccess-AI-Collective/axolotl.git
 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN cd axolotl && \
@@ -22,5 +21,10 @@ RUN cd axolotl && \
         pip install -e .; \
     fi
 
+# fix so that git fetch/pull from remote works
+RUN cd axolotl && \
+    git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
+    git config --get remote.origin.fetch
+
 # helper for huggingface-login cli
 RUN git config --global credential.helper store

docker/Dockerfile-base

@@ -8,7 +8,7 @@ FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION a
 ENV PATH="/root/miniconda3/bin:${PATH}"
 
 ARG PYTHON_VERSION="3.9"
-ARG PYTORCH="2.0.0"
+ARG PYTORCH_VERSION="2.0.1"
 ARG CUDA="118"
 
 ENV PYTHON_VERSION=$PYTHON_VERSION
@@ -29,17 +29,18 @@ ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
 WORKDIR /workspace
 
 RUN python3 -m pip install --upgrade pip && pip3 install packaging && \
-    python3 -m pip install --no-cache-dir -U torch==${PYTORCH} torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu$CUDA
+    python3 -m pip install --no-cache-dir -U torch==${PYTORCH_VERSION}+cu${CUDA} --extra-index-url https://download.pytorch.org/whl/cu$CUDA
 
 
 FROM base-builder AS flash-attn-builder
 
 WORKDIR /workspace
 
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
 
-RUN git clone https://github.com/HazyResearch/flash-attention.git && \
+RUN git clone https://github.com/Dao-AILab/flash-attention.git && \
     cd flash-attention && \
+    git checkout v2.0.1  && \
     python3 setup.py bdist_wheel && \
     cd csrc/fused_dense_lib && \
     python3 setup.py bdist_wheel && \
@@ -52,7 +53,7 @@ RUN git clone https://github.com/HazyResearch/flash-attention.git && \
 
 FROM base-builder AS deepspeed-builder
 
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
 
 WORKDIR /workspace
 
@@ -73,11 +74,14 @@ RUN git clone https://github.com/TimDettmers/bitsandbytes.git && \
 
 FROM base-builder
 
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
+ENV TORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST
+
 # recompile apex
 RUN python3 -m pip uninstall -y apex
 RUN git clone https://github.com/NVIDIA/apex
 #  `MAX_JOBS=1` disables parallel building to avoid cpu memory OOM when building image on GitHub Action (standard) runners
-RUN cd apex && MAX_JOBS=1 python3 -m pip install --global-option="--cpp_ext" --global-option="--cuda_ext" --no-cache -v --disable-pip-version-check .
+RUN cd apex && MAX_JOBS=1 python3 -m pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext" --config-settings "--build-option=--cuda_ext" ./
 
 RUN mkdir -p /workspace/builds
 COPY --from=bnb-builder /workspace/bitsandbytes /workspace/builds/bitsandbytes
@@ -97,4 +101,4 @@ RUN cd /workspace/builds/bitsandbytes && python3 setup.py install
 RUN git lfs install --skip-repo
 RUN pip3 install awscli && \
     # The base image ships with `pydantic==1.8.2` which is not working
-    pip3 install -U --no-cache-dir pydantic
+    pip3 install -U --no-cache-dir pydantic==1.10.10

docker/Dockerfile-runpod

@@ -1,6 +1,10 @@
 ARG BASE_TAG=main
 FROM winglian/axolotl:$BASE_TAG
 
+ENV HF_DATASETS_CACHE="/workspace/data/huggingface-cache/datasets"
+ENV HUGGINGFACE_HUB_CACHE="/workspace/data/huggingface-cache/hub"
+ENV TRANSFORMERS_CACHE="/workspace/data/huggingface-cache/hub"
+
 COPY scripts/runpod-entrypoint.sh /root/runpod-entrypoint.sh
 
 RUN apt install --yes --no-install-recommends openssh-server tmux && \

examples/llama-2/README.md

@@ -0,0 +1,20 @@
+# Overview
+
+This is an example of a llama-2 configuration for 7b and 13b. The yaml file contains configuration for the 7b variant, but you can just aswell use the same settings for 13b.
+
+The 7b variant fits on any 24GB VRAM GPU and will take up about 17 GB of VRAM during training if using qlora and 20 GB if using lora. On a RTX 4090 it trains 3 epochs of the default dataset in about 15 minutes.
+
+The 13b variant will fit if you change these settings to these values:
+gradient_accumulation_steps: 2
+micro_batch_size: 1
+
+```shell
+accelerate launch scripts/finetune.py examples/llama-2/qlora.yml
+
+```
+or
+
+```shell
+accelerate launch scripts/finetune.py examples/llama-2/lora.yml
+
+```

examples/llama-2/lora.yml

@@ -0,0 +1,66 @@
+base_model: meta-llama/Llama-2-7b-hf
+base_model_config: meta-llama/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.01
+output_dir: ./lora-out
+
+sequence_len: 4096
+max_packed_sequence_len: 4096
+
+adapter: lora
+lora_model_dir:
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 3
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: true
+bf16: true
+fp16: false
+tf32: false
+
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention: true
+flash_attention:
+
+warmup_steps: 10
+eval_steps: 20
+save_steps:
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"
+  pad_token: "<pad>"

examples/llama-2/qlora.yml

@@ -0,0 +1,67 @@
+base_model: meta-llama/Llama-2-7b-hf
+base_model_config: meta-llama/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.01
+output_dir: ./qlora-out
+
+adapter: qlora
+lora_model_dir:
+
+sequence_len: 4096
+max_packed_sequence_len: 4096
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules:
+lora_target_linear: true
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 3
+optimizer: paged_adamw_32bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: true
+bf16: true
+fp16: false
+tf32: false
+
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention: true
+flash_attention:
+
+warmup_steps: 10
+eval_steps: 20
+save_steps:
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"
+  pad_token: "<pad>"

examples/pythia-12b/README.md

@@ -0,0 +1,9 @@
+# Pythia 12B
+
+- Single-GPU A100 only (?)
+
+```shell
+python scripts/finetune.py examples/pythia-12b/config.yml
+```
+
+⚠️ Multiple-GPU A100 - Doesn't seem to work with multi-gpu without causing OOM! ⚠️

examples/pythia-12b/config.yml

@@ -0,0 +1,49 @@
+base_model: EleutherAI/pythia-12b-deduped
+base_model_config: EleutherAI/pythia-12b-deduped
+base_model_ignore_patterns: pytorch*  # prefer safetensors
+model_type: GPTNeoXForCausalLM
+tokenizer_type: AutoTokenizer
+load_in_8bit: false
+load_in_4bit: false
+gptq: false
+device_map: auto
+datasets:
+  - path: vicgalle/alpaca-gpt4
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+adapter:
+lora_model_dir:
+sequence_len: 2048
+max_packed_sequence_len: 2048
+lora_r: 64
+lora_alpha: 32
+lora_dropout: 0.0
+lora_target_modules:
+lora_target_linear: true
+lora_fan_in_fan_out: true  # pythia/GPTNeoX lora specific
+wandb_project:
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+output_dir: ./pythia-12b
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 5
+learning_rate: 0.00003
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+train_on_inputs: false
+group_by_length: false
+bf16: false
+fp16: false
+float16: true
+tf32: true
+flash_optimum: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+gradient_checkpointing: true
+fsdp:
+fsdp_config:
+collator_pad_to_longest: true

examples/redpajama/config-3b.yml

@@ -1,7 +1,7 @@
 base_model: togethercomputer/RedPajama-INCITE-Chat-3B-v1
 base_model_config: togethercomputer/RedPajama-INCITE-Chat-3B-v1
 model_type: GPTNeoXForCausalLM
-tokenizer_type: GPTNeoXTokenizer
+tokenizer_type: AutoTokenizer
 trust_remote_code:
 load_in_8bit: false
 datasets:

examples/xgen-7b/xgen-7b-8k-qlora.yml

@@ -0,0 +1,90 @@
+# An example finetuning Saleforce's XGen-7b model with 8k context using qlora
+# on Tim Dettmer's Guanaco dataset.
+base_model: Salesforce/xgen-7b-8k-base
+base_model_config: Salesforce/xgen-7b-8k-base
+trust_remote_code: true
+model_type: AutoModelForCausalLM
+tokenizer_type: AutoTokenizer
+load_in_8bit: false
+# enable 4bit for QLoRA
+load_in_4bit: true
+gptq: false
+strict: false
+push_dataset_to_hub:
+datasets:
+  - path: timdettmers/openassistant-guanaco
+    data_files:
+      - openassistant_best_replies_train.jsonl
+    type: "completion"
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.01
+# enable QLoRA
+adapter: qlora
+lora_model_dir:
+sequence_len: 8192
+max_packed_sequence_len:
+
+# hyperparameters from QLoRA paper Appendix B.2
+# "We find hyperparameters to be largely robust across datasets"
+lora_r: 64
+lora_alpha: 16
+# 0.1 for models up to 13B
+# 0.05 for 33B and 65B models
+lora_dropout: 0.05
+# add LoRA modules on all linear layers of the base model
+lora_target_modules:
+lora_target_linear: true
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+output_dir: ./qlora-out
+
+# QLoRA paper Table 9
+# - 16 for 7b & 13b
+# - 32 for 33b, 64 for 64b
+# Max size tested on A6000
+# - 7b: 40
+# - 40b: 4
+# decrease if OOM, increase for max VRAM utilization
+micro_batch_size: 1
+gradient_accumulation_steps: 1
+num_epochs: 3
+# Optimizer for QLoRA
+optimizer: paged_adamw_32bit
+torchdistx_path:
+lr_scheduler: cosine
+# QLoRA paper Table 9
+# - 2e-4 for 7b & 13b
+# - 1e-4 for 33b & 64b
+learning_rate: 0.00002
+train_on_inputs: false
+group_by_length: false
+bf16: true
+fp16: false
+tf32: false
+gradient_checkpointing: true
+# stop training after this many evaluation losses have increased in a row
+# https://huggingface.co/transformers/v4.2.2/_modules/transformers/trainer_callback.html#EarlyStoppingCallback
+early_stopping_patience: 3
+resume_from_checkpoint:
+auto_resume_from_checkpoints: true
+local_rank:
+logging_steps: 1
+xformers_attention: true
+flash_attention:
+gptq_groupsize:
+gptq_model_v1:
+warmup_steps: 10
+eval_steps: 50
+save_steps: 50
+debug:
+deepspeed:
+weight_decay: 0.0
+special_tokens:
+  eos_token: "<|endoftext|>"
+  bos_token: "<|endoftext|>"
+  unk_token: "<|endoftext|>"
+  pad_token: "<|endoftext|>"

requirements.txt

@@ -1,7 +1,7 @@
 peft @ git+https://github.com/huggingface/peft.git
 transformers @ git+https://github.com/huggingface/transformers.git
 bitsandbytes>=0.39.0
-accelerate
+accelerate @ git+https://github.com/huggingface/accelerate@2a289f6108e77a77a4efffb3f6316bc98538413b
 addict
 fire
 PyYAML==6.0
@@ -11,6 +11,8 @@ sentencepiece
 wandb
 einops
 xformers
+optimum
+hf_transfer
 # qlora things
 bert-score==0.3.13
 evaluate==0.4.0

scripts/alpaca_json_to_jsonl.py

@@ -15,6 +15,9 @@ from axolotl.convert import (
     JsonToJsonlConverter,
     StdoutWriter,
 )
+from axolotl.logging_config import configure_logging
+
+configure_logging()
 
 # add src to the pythonpath so we don't need to pip install this
 project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))

scripts/finetune.py

@@ -12,13 +12,15 @@ from typing import Any, Dict, List, Optional, Union
 import fire
 import torch
 import yaml
-from transformers import GenerationConfig, TextStreamer
-
-from axolotl.utils.data import load_prepare_datasets
-from axolotl.utils.dict import DictDefault
-from axolotl.utils.models import load_model, load_tokenizer
 
 # add src to the pythonpath so we don't need to pip install this
+from optimum.bettertransformer import BetterTransformer
+from transformers import GenerationConfig, TextStreamer
+
+from axolotl.logging_config import configure_logging
+from axolotl.utils.data import load_prepare_datasets, load_pretraining_dataset
+from axolotl.utils.dict import DictDefault
+from axolotl.utils.models import load_model, load_tokenizer
 from axolotl.utils.tokenization import check_dataset_labels
 from axolotl.utils.trainer import setup_trainer
 from axolotl.utils.validation import validate_config
@@ -28,9 +30,12 @@ project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
 src_dir = os.path.join(project_root, "src")
 sys.path.insert(0, src_dir)
 
+configure_logging()
+LOG = logging.getLogger("axolotl.scripts")
+
 
-logging.basicConfig(level=os.getenv("LOG_LEVEL", "INFO"))
 DEFAULT_DATASET_PREPARED_PATH = "last_run_prepared"
+os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
 
 
 def choose_device(cfg):
@@ -211,18 +216,29 @@ def train(
 
     # load the tokenizer first
     tokenizer_config = cfg.tokenizer_config or cfg.base_model_config
-    logging.info(f"loading tokenizer... {tokenizer_config}")
+    LOG.info(f"loading tokenizer... {tokenizer_config}")
     tokenizer = load_tokenizer(tokenizer_config, cfg.tokenizer_type, cfg)
 
     if (
         check_not_in(["shard", "merge_lora"], kwargs) and not cfg.inference
     ):  # don't need to load dataset for these
-        train_dataset, eval_dataset = load_prepare_datasets(
+        if not cfg.pretraining_dataset:
-            tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH
+            train_dataset, eval_dataset = load_prepare_datasets(
-        )
+                tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH
+            )
+        else:
+            train_dataset = load_pretraining_dataset(
+                cfg.pretraining_dataset,
+                tokenizer,
+                max_tokens=cfg.sequence_len,
+                seed=cfg.seed,
+            )
+            # https://discuss.huggingface.co/t/how-to-use-huggingface-trainer-streaming-datasets-without-wrapping-it-with-torchdatas-iterablewrapper/25230
+            train_dataset = train_dataset.with_format("torch")
+            eval_dataset = None
 
     if cfg.debug or "debug" in kwargs:
-        logging.info("check_dataset_labels...")
+        LOG.info("check_dataset_labels...")
         check_dataset_labels(
             train_dataset.select(
                 [random.randrange(0, len(train_dataset) - 1) for _ in range(5)]  # nosec
@@ -231,11 +247,11 @@ def train(
         )
 
     if prepare_ds_only:
-        logging.info("Finished preparing dataset. Exiting...")
+        LOG.info("Finished preparing dataset. Exiting...")
         return
 
     # Load the model and tokenizer
-    logging.info("loading model and peft_config...")
+    LOG.info("loading model and peft_config...")
     model, peft_config = load_model(
         cfg.base_model,
         cfg.base_model_config,
@@ -246,17 +262,17 @@ def train(
     )
 
     if "merge_lora" in kwargs and cfg.adapter is not None:
-        logging.info("running merge of LoRA with base model")
+        LOG.info("running merge of LoRA with base model")
         model = model.merge_and_unload()
         model.to(dtype=torch.float16)
 
         if cfg.local_rank == 0:
-            logging.info("saving merged model")
+            LOG.info("saving merged model")
             model.save_pretrained(str(Path(cfg.output_dir) / "merged"))
         return
 
     if cfg.inference:
-        logging.info("calling do_inference function")
+        LOG.info("calling do_inference function")
         prompter: Optional[str] = "AlpacaPrompter"
         if "prompter" in kwargs:
             if kwargs["prompter"] == "None":
@@ -275,27 +291,30 @@ def train(
     model.config.use_cache = False
 
     if torch.__version__ >= "2" and sys.platform != "win32":
-        logging.info("Compiling torch model")
+        LOG.info("Compiling torch model")
         model = torch.compile(model)
 
     # go ahead and presave, so we have the adapter config available to inspect
     if peft_config:
-        logging.info(f"Pre-saving adapter config to {cfg.output_dir}")
+        LOG.info(f"Pre-saving adapter config to {cfg.output_dir}")
         peft_config.save_pretrained(cfg.output_dir)
 
     # In case we want to stop early with ctrl+c, this is a nice to have to save the pretrained model
     if cfg.local_rank == 0:
+
+        def terminate_handler(_, __, model):
+            if cfg.flash_optimum:
+                model = BetterTransformer.reverse(model)
+            model.save_pretrained(cfg.output_dir)
+            sys.exit(0)
+
         signal.signal(
-            signal.SIGINT,
+            signal.SIGINT, lambda signum, frame: terminate_handler(signum, frame, model)
-            lambda signal, frame: (
-                model.save_pretrained(cfg.output_dir),
-                sys.exit(0),
-            ),
         )
 
-    logging.info("Starting trainer...")
+    LOG.info("Starting trainer...")
     if cfg.group_by_length:
-        logging.info("hang tight... sorting dataset for group_by_length")
+        LOG.info("hang tight... sorting dataset for group_by_length")
     resume_from_checkpoint = cfg.resume_from_checkpoint
     if cfg.resume_from_checkpoint is None and cfg.auto_resume_from_checkpoints:
         possible_checkpoints = [
@@ -307,19 +326,29 @@ def train(
                 key=lambda path: int(path.split("-")[-1]),
             )
             resume_from_checkpoint = sorted_paths[-1]
-            logging.info(
+            LOG.info(
                 f"Using Auto-resume functionality to start with checkpoint at {resume_from_checkpoint}"
             )
 
     if not Path(cfg.output_dir).is_dir():
         os.makedirs(cfg.output_dir, exist_ok=True)
-    trainer.train(resume_from_checkpoint=resume_from_checkpoint)
+    if cfg.flash_optimum:
+        with torch.backends.cuda.sdp_kernel(
+            enable_flash=True, enable_math=True, enable_mem_efficient=True
+        ):
+            trainer.train(resume_from_checkpoint=resume_from_checkpoint)
+    else:
+        trainer.train(resume_from_checkpoint=resume_from_checkpoint)
 
-    logging.info(f"Training Completed!!! Saving pre-trained model to {cfg.output_dir}")
+    LOG.info(f"Training Completed!!! Saving pre-trained model to {cfg.output_dir}")
 
     # TODO do we need this fix? https://huggingface.co/docs/accelerate/usage_guides/fsdp#saving-and-loading
     # only save on rank 0, otherwise it corrupts output on multi-GPU when multiple processes attempt to write the same file
-    if cfg.local_rank == 0:
+    if cfg.fsdp:
+        model.save_pretrained(cfg.output_dir)
+    elif cfg.local_rank == 0:
+        if cfg.flash_optimum:
+            model = BetterTransformer.reverse(model)
         model.save_pretrained(cfg.output_dir)
 
     # trainer.save_model(cfg.output_dir)  # TODO this may be needed for deepspeed to work? need to review another time

scripts/runpod-entrypoint.sh

@@ -1,10 +1,21 @@
 #!/bin/bash
 
-echo $PUBLIC_KEY >> ~/.ssh/authorized_keys
+# Export specific ENV variables to /etc/rp_environment
-chmod 700 -R ~/.ssh
+echo "Exporting environment variables..."
+printenv | grep -E '^RUNPOD_|^PATH=|^_=' | sed 's/^\(.*\)=\(.*\)$/export \1="\2"/' >> /etc/rp_environment
+echo 'source /etc/rp_environment' >> ~/.bashrc
 
-# Start the SSH service in the background
+if [[ $PUBLIC_KEY ]]
-service ssh start
+then
+    mkdir -p ~/.ssh
+    chmod 700 ~/.ssh
+    echo $PUBLIC_KEY >> ~/.ssh/authorized_keys
+    chmod 700 -R ~/.ssh
+    # Start the SSH service in the background
+    service ssh start
+else
+    echo "No PUBLIC_KEY ENV variable provided, not starting openSSH daemon"
+fi
 
 # Execute the passed arguments (CMD)
 exec "$@"

src/axolotl/datasets.py

@@ -1,12 +1,13 @@
 """Module containing Dataset functionality"""
 
 import logging
+import os
 from typing import List
 
 import torch
 from datasets import IterableDataset
 
-from .prompt_tokenizers import InvalidDataException, PromptTokenizingStrategy
+from .prompt_tokenizers import PromptTokenizingStrategy
 
 # We want this to be a wrapper for an existing dataset that we have loaded
 # lets use the concept of middlewares to wrap each dataset, for example
@@ -14,6 +15,8 @@ from .prompt_tokenizers import InvalidDataException, PromptTokenizingStrategy
 # let's check to ensure we don't truncate an item in the middle, we'll use
 # the collators later on to pad the datasets
 
+LOG = logging.getLogger("axolotl")
+
 
 class TokenizedPromptDataset(IterableDataset):
     """
@@ -32,17 +35,15 @@ class TokenizedPromptDataset(IterableDataset):
         self.dataset = dataset
 
     def __iter__(self):
-        iterator = iter(self.dataset)
+        features = self.dataset.features.keys()
-        count = 0
+        num_proc = os.cpu_count()
-        # Loop through the entire dataset
+        return iter(
-        for example in iterator:
+            self.dataset.map(
-            try:
+                self.prompt_tokenizer.tokenize_prompt,
-                yield self.prompt_tokenizer.tokenize_prompt(example)
+                num_proc=num_proc,
-                count += 1
+                remove_columns=features,
-            except InvalidDataException:
+            )
-                pass
+        )
-        if count == 0:
-            raise RuntimeError("Expected at least one datapoint in dataset.")
 
 
 # TODO this isn't the best since it can't interleave datasets
@@ -115,7 +116,7 @@ class ConstantLengthDataset(IterableDataset):
                                 "attention_mask": attention_mask,
                             }
                         else:
-                            logging.warning(
+                            LOG.warning(
                                 f"dropping batch due to tensor size mismatch input_ids: {input_ids.size()}, labels: {labels.size()}, attention_mask: {attention_mask.size()}"
                             )
                     buffer = {
@@ -126,6 +127,7 @@ class ConstantLengthDataset(IterableDataset):
                     buffer_len = 0
 
                 if example:
+                    # FIXME
                     # just going to drop data points that are too long
                     if len(example["input_ids"]) <= self.seq_length:
                         input_ids = example["input_ids"]

src/axolotl/logging_config.py

@@ -0,0 +1,33 @@
+"""Logging configuration settings"""
+
+import os
+import sys
+from logging.config import dictConfig
+from typing import Any, Dict
+
+DEFAULT_LOGGING_CONFIG: Dict[str, Any] = {
+    "version": 1,
+    "formatters": {
+        "simple": {
+            "format": "[%(asctime)s] [%(levelname)s] [%(name)s.%(funcName)s:%(lineno)d] [PID:%(process)d] %(message)s",
+        },
+    },
+    "filters": {},
+    "handlers": {
+        "console": {
+            "class": "logging.StreamHandler",
+            "formatter": "simple",
+            "filters": [],
+            "stream": sys.stdout,
+        },
+    },
+    "root": {"handlers": ["console"], "level": os.getenv("LOG_LEVEL", "INFO")},
+    "loggers": {
+        "axolotl": {"handlers": ["console"], "level": "DEBUG", "propagate": False},
+    },
+}
+
+
+def configure_logging():
+    """Configure with default logging"""
+    dictConfig(DEFAULT_LOGGING_CONFIG)

src/axolotl/monkeypatch/llama_attn_hijack_flash.py

@@ -8,7 +8,7 @@ import torch
 import transformers
 from einops import rearrange
 from flash_attn.bert_padding import pad_input, unpad_input
-from flash_attn.flash_attn_interface import flash_attn_unpadded_qkvpacked_func
+from flash_attn.flash_attn_interface import flash_attn_varlen_qkvpacked_func
 from transformers.models.llama.modeling_llama import apply_rotary_pos_emb
 
 
@@ -79,7 +79,7 @@ def forward(
             dtype=torch.int32,
             device=qkv.device,
         )
-        output = flash_attn_unpadded_qkvpacked_func(
+        output = flash_attn_varlen_qkvpacked_func(
             qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True
         )
         output = rearrange(output, "(b s) ... -> b s ...", b=bsz)
@@ -95,7 +95,7 @@ def forward(
             three=3,
             h=nheads,
         )
-        output_unpad = flash_attn_unpadded_qkvpacked_func(
+        output_unpad = flash_attn_varlen_qkvpacked_func(
             x_unpad,
             cu_q_lens,
             max_s,

src/axolotl/monkeypatch/llama_attn_hijack_xformers.py

@@ -7,6 +7,7 @@ import math
 from typing import Optional, Tuple
 
 import torch
+import torch.nn.functional as F
 import transformers.models.llama.modeling_llama
 from torch import nn
 
@@ -38,21 +39,48 @@ def xformers_forward(
     # pylint: disable=duplicate-code
     bsz, q_len, _ = hidden_states.size()
 
-    query_states = (
+    if not hasattr(self, "pretraining_tp"):
-        self.q_proj(hidden_states)
+        self.pretraining_tp = 1
-        .view(bsz, q_len, self.num_heads, self.head_dim)
+
-        .transpose(1, 2)
+    if self.pretraining_tp > 1:
-    )
+        key_value_slicing = (
-    key_states = (
+            self.num_key_value_heads * self.head_dim
-        self.k_proj(hidden_states)
+        ) // self.pretraining_tp
-        .view(bsz, q_len, self.num_heads, self.head_dim)
+        query_slices = self.q_proj.weight.split(
-        .transpose(1, 2)
+            (self.num_heads * self.head_dim) // self.pretraining_tp, dim=0
-    )
+        )
-    value_states = (
+        key_slices = self.k_proj.weight.split(key_value_slicing, dim=0)
-        self.v_proj(hidden_states)
+        value_slices = self.v_proj.weight.split(key_value_slicing, dim=0)
-        .view(bsz, q_len, self.num_heads, self.head_dim)
+
-        .transpose(1, 2)
+        query_states = [
-    )
+            F.linear(hidden_states, query_slices[i]) for i in range(self.pretraining_tp)
+        ]
+        query_states = torch.cat(query_states, dim=-1)
+
+        key_states = [
+            F.linear(hidden_states, key_slices[i]) for i in range(self.pretraining_tp)
+        ]
+        key_states = torch.cat(key_states, dim=-1)
+
+        value_states = [
+            F.linear(hidden_states, value_slices[i]) for i in range(self.pretraining_tp)
+        ]
+        value_states = torch.cat(value_states, dim=-1)
+
+    else:
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+    query_states = query_states.view(
+        bsz, q_len, self.num_heads, self.head_dim
+    ).transpose(1, 2)
+    key_states = key_states.view(
+        bsz, q_len, self.num_key_value_heads, self.head_dim
+    ).transpose(1, 2)
+    value_states = value_states.view(
+        bsz, q_len, self.num_key_value_heads, self.head_dim
+    ).transpose(1, 2)
 
     kv_seq_len = key_states.shape[-2]
     if past_key_value is not None:
@@ -73,6 +101,14 @@ def xformers_forward(
 
     past_key_value = (key_states, value_states) if use_cache else None
 
+    # repeat k/v heads if n_kv_heads < n_heads
+    key_states = transformers.models.llama.modeling_llama.repeat_kv(
+        key_states, self.num_key_value_groups
+    )
+    value_states = transformers.models.llama.modeling_llama.repeat_kv(
+        value_states, self.num_key_value_groups
+    )
+
     # We only apply xformers optimizations if we don't need to output the whole attention matrix
     if not output_attentions:
         query_states = query_states.transpose(1, 2)
@@ -128,10 +164,23 @@ def xformers_forward(
                 f" {attn_output.size()}"
             )
 
-        attn_output = attn_output.transpose(1, 2)
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        # end x-formers vs. not x-formers if-else block
 
     attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
-    attn_output = self.o_proj(attn_output)
+
+    if self.pretraining_tp > 1:
+        attn_output = attn_output.split(self.hidden_size // self.pretraining_tp, dim=2)
+        o_proj_slices = self.o_proj.weight.split(
+            self.hidden_size // self.pretraining_tp, dim=1
+        )
+        attn_output = sum(
+            F.linear(attn_output[i], o_proj_slices[i])
+            for i in range(self.pretraining_tp)
+        )
+    else:
+        attn_output = self.o_proj(attn_output)
+
     return attn_output, attn_weights, past_key_value
 
 
@@ -184,14 +233,15 @@ def sdp_attention_forward(
 
     # We only apply sdp attention if we don't need to output the whole attention matrix
     if not output_attentions:
-        attn_output = torch.nn.functional.scaled_dot_product_attention(
+        with torch.backends.cuda.sdp_kernel():
-            query_states,
+            attn_output = torch.nn.functional.scaled_dot_product_attention(
-            key_states,
+                query_states,
-            value_states,
+                key_states,
-            attn_mask=attention_mask,
+                value_states,
-            is_causal=False,
+                attn_mask=attention_mask,
-        )
+                is_causal=False,
-        attn_weights = None
+            )
+            attn_weights = None
     else:
         attn_weights = torch.matmul(
             query_states, key_states.transpose(2, 3)

src/axolotl/monkeypatch/llama_landmark_attn.py

@@ -53,7 +53,7 @@ from transformers.utils import (
     replace_return_docstrings,
 )
 
-logger = logging.get_logger(__name__)
+LOG = logging.getLogger("axolotl")
 
 _CONFIG_FOR_DOC = "LlamaConfig"
 
@@ -862,7 +862,7 @@ class LlamaModel(LlamaPreTrainedModel):
 
         if self.gradient_checkpointing and self.training:
             if use_cache:
-                logger.warning_once(
+                LOG.warning_once(
                     "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                 )
                 use_cache = False

src/axolotl/prompt_strategies/alpaca_chat.py

@@ -6,7 +6,7 @@ from axolotl.prompt_tokenizers import (
     AlpacaPromptTokenizingStrategy,
     InstructionPromptTokenizingStrategy,
 )
-from axolotl.prompters import AlpacaPrompter, PromptStyle
+from axolotl.prompters import AlpacaPrompter, PromptStyle, UnpromptedPrompter
 
 
 def load(tokenizer, cfg):
@@ -20,11 +20,38 @@ def load(tokenizer, cfg):
 
 class AlpacaConcisePrompter(AlpacaPrompter):
     """
-    Alpaca Prompter extending the system prompt to ask for concise answers
+    Alpaca Prompter extending the system prompt to ask for concise chat-instruct answers
     """
 
-    system_prompt = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that concisely and appropriately completes the request.\n\n"
+    system_prompt = "Below is an instruction from a USER that describes a task, paired with an input that provides further context. The ASSISTANT writes a response that concisely and appropriately completes the request.\n\n"
-    system_no_input_prompt = "Below is an instruction that describes a task. Write a response that appropriately and concisely completes the request.\n\n"
+    system_no_input_prompt = "Below is an instruction from a USER that describes a task. The ASSISTANT writes a response that appropriately and concisely completes the request.\n\n"
+
+
+class AlpacaChatPrompter(AlpacaPrompter):
+    """
+    Alpaca Chat Prompter extending the system prompt to for chat-instruct answers
+    """
+
+    system_prompt = "Below is an instruction from a USER that describes a task, paired with an input that provides further context. The ASSISTANT writes a response that concisely and appropriately completes the request.\n\n"
+    system_no_input_prompt = "Below is an instruction from a USER that describes a task. The ASSISTANT writes a response that appropriately and concisely completes the request.\n\n"
+
+    def __init__(self):  # pylint: disable=super-init-not-called
+        self.prompt_style = PromptStyle.CHAT.value
+        self.match_prompt_style()
+
+
+class NoSystemPrompter(AlpacaPrompter):
+    """
+    Null Prompter with no system prompts
+    """
+
+    system_prompt = ""
+    system_no_input_prompt = ""
+    turn_format = "{instruction} {input} "
+    turn_no_input_format = "{instruction} "
+
+    def __init__(self):  # pylint: disable=super-init-not-called
+        pass
 
 
 class AlpacaQAPromptTokenizingStrategy(InstructionPromptTokenizingStrategy):
@@ -64,7 +91,7 @@ def load_concise(tokenizer, cfg):
 
 def load_qa(tokenizer, cfg):
     return AlpacaQAPromptTokenizingStrategy(
-        AlpacaPrompter(PromptStyle.CHAT.value),
+        AlpacaChatPrompter(),
         tokenizer,
         cfg.train_on_inputs,
         cfg.sequence_len,
@@ -73,7 +100,16 @@ def load_qa(tokenizer, cfg):
 
 def load_camel_ai(tokenizer, cfg):
     return CamelAIPromptTokenizingStrategy(
-        AlpacaPrompter(PromptStyle.CHAT.value),
+        AlpacaChatPrompter(),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+
+
+def load_no_prompt(tokenizer, cfg):
+    return AlpacaPromptTokenizingStrategy(
+        UnpromptedPrompter(PromptStyle.CHAT.value),
         tokenizer,
         cfg.train_on_inputs,
         cfg.sequence_len,

src/axolotl/prompt_strategies/alpaca_instruct.py

@@ -1,7 +1,7 @@
 """Module loading the AlpacaInstructPromptTokenizingStrategy class"""
 
 from axolotl.prompt_tokenizers import AlpacaPromptTokenizingStrategy
-from axolotl.prompters import AlpacaPrompter, PromptStyle
+from axolotl.prompters import AlpacaPrompter, PromptStyle, UnpromptedPrompter
 
 
 def load(tokenizer, cfg):
@@ -11,3 +11,12 @@ def load(tokenizer, cfg):
         cfg.train_on_inputs,
         cfg.sequence_len,
     )
+
+
+def load_no_prompt(tokenizer, cfg):
+    return AlpacaPromptTokenizingStrategy(
+        UnpromptedPrompter(PromptStyle.INSTRUCT.value),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )

src/axolotl/prompt_strategies/alpaca_w_system.py

@@ -0,0 +1,139 @@
+"""
+Prompt strategies loader for alpaca instruction datasets with system prompts
+"""
+from typing import Generator, Tuple, Union
+
+from axolotl.prompt_tokenizers import PromptTokenizingStrategy
+from axolotl.prompters import AlpacaPrompter, PromptStyle
+
+
+class InstructionWSystemPromptTokenizingStrategy(PromptTokenizingStrategy):
+    """
+    Tokenizing strategy for instruction-based prompts.
+    """
+
+    def parse_instruction_fields(self, prompt) -> Tuple[str, str, str, str]:
+        return (
+            prompt["instruction"],
+            prompt["input"] if "input" in prompt else "",
+            prompt["output"],
+            prompt["system"],
+        )
+
+    def tokenize_prompt(self, prompt):
+        # pylint: disable=duplicate-code
+        (
+            instruction,
+            input,  # pylint: disable=redefined-builtin
+            response,
+            system,
+        ) = self.parse_instruction_fields(prompt)
+        user_prompt = next(
+            iter(
+                self.prompter.build_prompt_w_system(
+                    system,
+                    instruction,
+                    input,
+                )
+            )
+        )
+        tokenized_prompt = self._tokenize(user_prompt, add_eos_token=False)
+        if not self.train_on_inputs:
+            user_prompt_len = len(tokenized_prompt["input_ids"])
+            # TODO this could be sped up using numpy array slicing
+            tokenized_prompt["labels"] = [-100] * user_prompt_len
+        tokenized_res_prompt = self._tokenize(
+            response, strip_bos_token=True, add_eos_token=True
+        )
+        tokenized_prompt["input_ids"] += tokenized_res_prompt["input_ids"]
+        tokenized_prompt["attention_mask"] += tokenized_res_prompt["attention_mask"]
+        tokenized_prompt["labels"] += tokenized_res_prompt["input_ids"]
+
+        return tokenized_prompt
+
+
+class SystemDataPrompter(AlpacaPrompter):
+    """
+    Alpaca Style Prompter that uses system prompts from the dataset
+    """
+
+    def build_prompt_w_system(
+        self,
+        system: str,
+        instruction: str,
+        input: Union[None, str] = None,  # pylint: disable=redefined-builtin
+        output: Union[None, str] = None,
+    ) -> Generator[str, None, None]:
+        # returns the full prompt from instruction and optional input
+        # if a label (=response, =output) is provided, it's also appended.
+        formatted_sys_prompt = f"### System:\n{system}\n\n" if system else ""
+        if input:
+            res = formatted_sys_prompt + self.turn_format.format(
+                instruction=instruction, input=input
+            )
+        else:
+            res = formatted_sys_prompt + self.turn_no_input_format.format(
+                instruction=instruction
+            )
+        if output:
+            res = f"{res}{output}"
+        yield res
+
+
+class OpenOrcaSystemDataPrompter(SystemDataPrompter):
+    """
+    Alpaca Style Prompter that uses system prompts from the dataset, with OpenOrca prompts
+    """
+
+    def match_prompt_style(self):
+        if self.prompt_style == PromptStyle.INSTRUCT.value:
+            self.turn_format = "### User:\n{instruction}\n\n### Additional Context:\n{input}\n\n### Assistant:\n"
+            self.turn_no_input_format = "### User:\n{instruction}\n\n### Assistant:\n"
+        if self.prompt_style == PromptStyle.CHAT.value:
+            self.turn_format = "USER: {instruction}\n{input}\nASSISTANT:"
+            self.turn_no_input_format = "USER: {instruction}\nASSISTANT:"
+
+
+class OpenOrcaPromptTokenizingStrategy(InstructionWSystemPromptTokenizingStrategy):
+    """
+    Tokenizing strategy for OpenOrca datasets
+    """
+
+    def parse_instruction_fields(self, prompt) -> Tuple[str, str, str, str]:
+        return (
+            prompt["question"],
+            "",
+            prompt["response"],
+            prompt["system_prompt"],
+        )
+
+
+def load(tokenizer, cfg):
+    return load_chat(tokenizer, cfg)
+
+
+def load_instruct(tokenizer, cfg):
+    return InstructionWSystemPromptTokenizingStrategy(
+        SystemDataPrompter(PromptStyle.INSTRUCT.value),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+
+
+def load_chat(tokenizer, cfg):
+    return InstructionWSystemPromptTokenizingStrategy(
+        SystemDataPrompter(PromptStyle.CHAT.value),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )
+
+
+def load_open_orca(tokenizer, cfg):
+    return OpenOrcaPromptTokenizingStrategy(
+        OpenOrcaSystemDataPrompter(PromptStyle.INSTRUCT.value),
+        tokenizer,
+        cfg.train_on_inputs,
+        cfg.sequence_len,
+    )

src/axolotl/prompt_strategies/pygmalion.py

@@ -11,6 +11,8 @@ from axolotl.prompt_tokenizers import (
     tokenize_prompt_default,
 )
 
+LOG = logging.getLogger("axolotl")
+
 IGNORE_TOKEN_ID = -100
 
 
@@ -64,7 +66,7 @@ class PygmalionPromptTokenizingStrategy(PromptTokenizingStrategy):
                     *copy.deepcopy(res["input_ids"])
                 ][len(self.bot_prefix_token_ids) :]
             else:
-                logging.warning(f"unknown role in conversation: {role}")
+                LOG.warning(f"unknown role in conversation: {role}")
                 res = defaultdict(lambda: [])
 
             # pylint: disable=duplicate-code

src/axolotl/prompt_tokenizers.py

@@ -10,6 +10,8 @@ from transformers import PreTrainedTokenizer
 
 from axolotl.prompters import IGNORE_TOKEN_ID
 
+LOG = logging.getLogger("axolotl")
+
 IGNORE_INDEX = -100
 LLAMA_DEFAULT_PAD_TOKEN = "[PAD]"  # nosec
 LLAMA_DEFAULT_EOS_TOKEN = "</s>"  # nosec
@@ -46,16 +48,22 @@ class PromptTokenizingStrategy(abc.ABC):
 
     @functools.lru_cache(maxsize=128)
     def _get_user_token(self):
-        id_or_ids = self.tokenizer.convert_tokens_to_ids("<|USER|>")
+        try:
-        if isinstance(id_or_ids, (int,)):
+            id_or_ids = self.tokenizer.convert_tokens_to_ids("<|USER|>")
-            return id_or_ids
+            if isinstance(id_or_ids, (int,)):
+                return id_or_ids
+        except KeyError:
+            pass
         return False
 
     @functools.lru_cache(maxsize=128)
     def _get_assistant_token(self):
-        id_or_ids = self.tokenizer.convert_tokens_to_ids("<|ASSISTANT|>")
+        try:
-        if isinstance(id_or_ids, (int,)):
+            id_or_ids = self.tokenizer.convert_tokens_to_ids("<|ASSISTANT|>")
-            return id_or_ids
+            if isinstance(id_or_ids, (int,)):
+                return id_or_ids
+        except KeyError:
+            pass
         return False
 
     def _tokenize(self, prompt: str, add_eos_token=True, strip_bos_token=False):
@@ -87,7 +95,9 @@ class InstructionPromptTokenizingStrategy(PromptTokenizingStrategy):
     Tokenizing strategy for instruction-based prompts.
     """
 
-    def parse_instruction_fields(self, prompt) -> Tuple[str, str, str]:
+    def parse_instruction_fields(
+        self, prompt
+    ) -> Union[Tuple[str, str, str], Tuple[str, str, str, str]]:
         raise NotImplementedError
 
     def tokenize_prompt(self, prompt):
@@ -96,25 +106,27 @@ class InstructionPromptTokenizingStrategy(PromptTokenizingStrategy):
             input,  # pylint: disable=redefined-builtin
             response,
         ) = self.parse_instruction_fields(prompt)
-        full_prompt = self._build_full_prompt(instruction, input, response)
+        user_prompt = next(
-        tokenized_full_prompt = self._tokenize(full_prompt)
+            iter(
-        if not self.train_on_inputs:
+                self.prompter.build_prompt(
-            user_prompt = next(
+                    instruction,
-                iter(
+                    input,
-                    self.prompter.build_prompt(
-                        instruction,
-                        input,
-                    )
                 )
             )
-            tokenized_user_prompt = self._tokenize(user_prompt, add_eos_token=False)
+        )
-            user_prompt_len = len(tokenized_user_prompt["input_ids"])
+        tokenized_prompt = self._tokenize(user_prompt, add_eos_token=False)
+        if not self.train_on_inputs:
+            user_prompt_len = len(tokenized_prompt["input_ids"])
             # TODO this could be sped up using numpy array slicing
-            tokenized_full_prompt["labels"] = [
+            tokenized_prompt["labels"] = [-100] * user_prompt_len
-                -100
+        tokenized_res_prompt = self._tokenize(
-            ] * user_prompt_len + tokenized_full_prompt["labels"][user_prompt_len:]
+            response, strip_bos_token=True, add_eos_token=True
+        )
+        tokenized_prompt["input_ids"] += tokenized_res_prompt["input_ids"]
+        tokenized_prompt["attention_mask"] += tokenized_res_prompt["attention_mask"]
+        tokenized_prompt["labels"] += tokenized_res_prompt["input_ids"]
 
-        return tokenized_full_prompt
+        return tokenized_prompt
 
     def _build_full_prompt(
         self, instruction, input, response  # pylint: disable=redefined-builtin
@@ -380,7 +392,7 @@ class ShareGPTPromptTokenizingStrategy(PromptTokenizingStrategy):
                         # everything from this is masked out from the labels
                         labels = [IGNORE_TOKEN_ID] * len(res["input_ids"])
                     else:
-                        logging.warning(f"unhandled role: {part[0]}")
+                        LOG.warning(f"unhandled role: {part[0]}")
 
                 # pylint: disable=duplicate-code
                 result, current_len = parse_tokenized_to_result(
@@ -436,7 +448,7 @@ def parse_tokenized_to_result(
     result: Dict[str, List[int]],
     current_len: int,
     res: Dict[str, List[int]],
-    labels: list[int],
+    labels: List[int],
     pad_token_id: Union[int, None] = None,
 ) -> Tuple[Dict[str, List[int]], int]:
     """

src/axolotl/prompters.py

@@ -5,6 +5,7 @@ import logging
 from enum import Enum, auto
 from typing import Generator, List, Optional, Tuple, Union
 
+LOG = logging.getLogger("axolotl")
 IGNORE_TOKEN_ID = -100
 
 
@@ -24,6 +25,8 @@ class AlpacaPrompter:
 
     system_prompt = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n\n"
     system_no_input_prompt = "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n"
+    turn_format: str
+    turn_no_input_format: str
     prompt_style: Optional[PromptStyle] = None
 
     def __init__(self, prompt_style=PromptStyle.INSTRUCT.value):
@@ -32,23 +35,13 @@ class AlpacaPrompter:
 
     def match_prompt_style(self):
         if self.prompt_style == PromptStyle.INSTRUCT.value:
-            self.prompt_input = (
+            self.turn_format = "### Instruction:\n{instruction}\n\n### Input:\n{input}\n\n### Response:\n"
-                self.system_prompt
+            self.turn_no_input_format = (
-                + "### Instruction:\n{instruction}\n\n### Input:\n{input}\n\n### Response:\n"
+                "### Instruction:\n{instruction}\n\n### Response:\n"
             )
-            self.prompt_no_input = (
-                self.system_no_input_prompt
-                + "### Instruction:\n{instruction}\n\n### Response:\n"
-            )
-            self.response_split = "### Response:"
         if self.prompt_style == PromptStyle.CHAT.value:
-            self.prompt_input = (
+            self.turn_format = "USER: {instruction}\n{input}\nASSISTANT:"
-                self.system_prompt + "USER: {instruction}\n{input}\nASSISTANT:"
+            self.turn_no_input_format = "USER: {instruction}\nASSISTANT:"
-            )
-            self.prompt_no_input = (
-                self.system_no_input_prompt + "USER: {instruction}\nASSISTANT:"
-            )
-            self.response_split = "ASSISTANT:"
 
     def build_prompt(
         self,
@@ -59,16 +52,17 @@ class AlpacaPrompter:
         # returns the full prompt from instruction and optional input
         # if a label (=response, =output) is provided, it's also appended.
         if input:
-            res = self.prompt_input.format(instruction=instruction, input=input)
+            res = self.system_prompt + self.turn_format.format(
+                instruction=instruction, input=input
+            )
         else:
-            res = self.prompt_no_input.format(instruction=instruction)
+            res = self.system_no_input_prompt + self.turn_no_input_format.format(
+                instruction=instruction
+            )
         if output:
             res = f"{res}{output}"
         yield res
 
-    def get_response(self, output: str) -> str:
-        return output.split(self.response_split)[1].strip()
-
 
 class UnpromptedPrompter(AlpacaPrompter):
     """
@@ -93,7 +87,10 @@ class MultipleChoiceExplainPrompter(AlpacaPrompter):
     """
 
     system_prompt = (
-        "Choose the answer that best answers the question. Explain your reasoning."
+        "Choose the answer that best answers the question. Explain your reasoning.\n"
+    )
+    system_no_input_prompt = (
+        "Choose the answer that best answers the question. Explain your reasoning.\n"
     )
 
 
@@ -102,7 +99,12 @@ class MultipleChoiceConcisePrompter(AlpacaPrompter):
     Prompter for multiple choice concise
     """
 
-    prompt_input = "Choose the answer that best answers the question. Be concise in your response.\n\nUSER: {instruction}\n{input}\nASSISTANT:\n"
+    system_prompt = "Choose the answer that best answers the question. Be concise in your response.\n\n"
+    system_no_input_prompt = "Choose the answer that best answers the question. Be concise in your response.\n\n"
+
+    def match_prompt_style(self):
+        self.turn_format = "USER: {instruction}\n{input}\nASSISTANT:"
+        self.turn_no_input_format = "USER: {instruction}\nASSISTANT:"
 
 
 class SummarizeTLDRPrompter(AlpacaPrompter):
@@ -110,9 +112,12 @@ class SummarizeTLDRPrompter(AlpacaPrompter):
     Prompter for summarize TLDR
     """
 
-    prompt_no_input = (
+    system_prompt = ""
-        "USER: Summarize the following article as a TL;DR.\n{instruction}\nASSISTANT:"
+    system_no_input_prompt = ""
-    )
+
+    def match_prompt_style(self):
+        self.turn_format = "USER: Summarize the following article as a TL;DR.\n{instruction}\n{input}\nASSISTANT:"
+        self.turn_no_input_format = "USER: Summarize the following article as a TL;DR.\n{instruction}\nASSISTANT:"
 
 
 class CompletionPrompter:
@@ -128,9 +133,6 @@ class CompletionPrompter:
     ) -> Generator[str, None, None]:
         yield instruction
 
-    def get_response(self, output: str) -> str:
-        return output.strip()
-
 
 class GPTeacherPrompter(AlpacaPrompter):
     """
@@ -210,9 +212,6 @@ class ReflectAlpacaPrompter:
             res = f"{res}{label}"
         yield res
 
-    def get_response(self, output: str) -> str:
-        return output.split(self.response_split)[1].strip()
-
 
 class SeparatorStyle(Enum):
     """Different separator style."""
@@ -243,7 +242,7 @@ class Conversation:
             if message:
                 yield (role + ":", " " + message)
             else:
-                logging.warning(f"role with empty message: {role}")
+                LOG.warning(f"role with empty message: {role}")
                 yield (role + ":", "")
 
     def copy(self):
@@ -289,12 +288,6 @@ class ShareGPTPrompter:  # pylint: disable=too-few-public-methods
             sep2=" ",
         )
 
-    # def match_prompt_style(self):
-    #     if self.prompt_style == PromptStyle.chat.value:
-    #         self.prompt_input = self.system_prompt + "USER: {instruction}\n{input}\nASSISTANT:"
-    #         self.prompt_no_input = self.system_no_input_prompt + "USER: {instruction}\nASSISTANT:"
-    #         self.response_split = "ASSISTANT:"
-
     def build_prompt(self, source) -> Generator[str, None, None]:
         # ignore the system prompt if provided
         if source[0]["from"] == "system":

src/axolotl/utils/callbacks.py

@@ -2,13 +2,14 @@
 
 import os
 
+from optimum.bettertransformer import BetterTransformer
 from transformers import (
     TrainerCallback,
     TrainerControl,
     TrainerState,
     TrainingArguments,
 )
-from transformers.trainer_utils import PREFIX_CHECKPOINT_DIR
+from transformers.trainer_utils import PREFIX_CHECKPOINT_DIR, IntervalStrategy
 
 
 class SavePeftModelCallback(TrainerCallback):  # pylint: disable=too-few-public-methods
@@ -30,3 +31,39 @@ class SavePeftModelCallback(TrainerCallback):  # pylint: disable=too-few-public-
         kwargs["model"].save_pretrained(peft_model_path)
 
         return control
+
+
+class SaveBetterTransformerModelCallback(
+    TrainerCallback
+):  # pylint: disable=too-few-public-methods
+    """Callback to save the BetterTransformer wrapped model"""
+
+    def on_step_end(
+        self,
+        args: TrainingArguments,
+        state: TrainerState,
+        control: TrainerControl,
+        **kwargs,
+    ):
+        # Save
+        if (
+            args.save_strategy == IntervalStrategy.STEPS
+            and args.save_steps > 0
+            and state.global_step % args.save_steps == 0
+        ):
+            control.should_save = True
+
+        if control.should_save:
+            checkpoint_folder = os.path.join(
+                args.output_dir,
+                f"{PREFIX_CHECKPOINT_DIR}-{state.global_step}",
+            )
+
+            model = BetterTransformer.reverse(kwargs["model"])
+            model.save_pretrained(checkpoint_folder)
+            # FIXME - need to cleanup old checkpoints
+
+            # since we're saving here, we don't need the trainer loop to attempt to save too b/c
+            # the trainer will raise an exception since it can't save a BetterTransformer wrapped model
+            control.should_save = False
+        return control

src/axolotl/utils/data.py

@@ -1,10 +1,12 @@
 """Module containing data utilities"""
-
+import functools
+import itertools
 import logging
 from hashlib import md5
 from pathlib import Path
 from typing import List, Tuple, Union
 
+import torch
 from datasets import Dataset, DatasetDict, load_dataset, load_from_disk
 from huggingface_hub import hf_hub_download
 from transformers import PreTrainedTokenizerBase
@@ -34,6 +36,8 @@ from axolotl.prompters import (
     SummarizeTLDRPrompter,
 )
 
+LOG = logging.getLogger("axolotl")
+
 
 def load_tokenized_prepared_datasets(
     tokenizer, cfg, default_dataset_prepared_path
@@ -72,17 +76,17 @@ def load_tokenized_prepared_datasets(
     if dataset:
         ...
     elif any(prepared_ds_path.glob("*")):
-        logging.info(f"Loading prepared dataset from disk at {prepared_ds_path}...")
+        LOG.info(f"Loading prepared dataset from disk at {prepared_ds_path}...")
         dataset = load_from_disk(str(prepared_ds_path))
-        logging.info("Prepared dataset loaded from disk...")
+        LOG.info("Prepared dataset loaded from disk...")
     else:
-        logging.info(f"Unable to find prepared dataset in {prepared_ds_path}")
+        LOG.info(f"Unable to find prepared dataset in {prepared_ds_path}")
-        logging.info("Loading raw datasets...")
+        LOG.info("Loading raw datasets...")
 
         if cfg.seed:
             seed = cfg.seed
         else:
-            logging.info("No seed provided, using default seed of 42")
+            LOG.info("No seed provided, using default seed of 42")
             seed = 42
 
         datasets = []
@@ -93,6 +97,7 @@ def load_tokenized_prepared_datasets(
             try:
                 load_dataset(
                     d.path,
+                    name=d.name,
                     streaming=True,
                     use_auth_token=use_auth_token,
                 )
@@ -101,34 +106,45 @@ def load_tokenized_prepared_datasets(
                 pass
 
             # prefer local dataset, even if hub exists
-            if Path(d.path).exists():
+            local_path = Path(d.path)
-                ds = load_dataset(
+            if local_path.exists():
-                    "json",
+                if local_path.is_dir():
-                    data_files=d.path,
-                    streaming=False,
-                    split=None,
-                )
-            elif ds_from_hub:
-                if d.data_files:
                     ds = load_dataset(
                         d.path,
-                        streaming=False,
+                        name=d.name,
                         data_files=d.data_files,
-                        use_auth_token=use_auth_token,
+                        streaming=False,
+                        split=None,
+                    )
+                elif local_path.is_file():
+                    ds = load_dataset(
+                        "json",
+                        name=d.name,
+                        data_files=d.path,
+                        streaming=False,
+                        split=None,
                     )
                 else:
-                    ds = load_dataset(
+                    raise ValueError(
-                        d.path,
+                        "unhandled dataset load: local path exists, but is neither a directory or a file"
-                        streaming=False,
-                        use_auth_token=use_auth_token,
                     )
+            elif ds_from_hub:
+                ds = load_dataset(
+                    d.path,
+                    name=d.name,
+                    streaming=False,
+                    data_files=d.data_files,
+                    use_auth_token=use_auth_token,
+                )
             else:
                 fp = hf_hub_download(
                     repo_id=d.path,
                     repo_type="dataset",
                     filename=d.data_files,
                 )
-                ds = load_dataset("json", data_files=fp, streaming=False, split=None)
+                ds = load_dataset(
+                    "json", name=d.name, data_files=fp, streaming=False, split=None
+                )
             if not ds:
                 raise ValueError("unhandled dataset load")
             # support for using a subset of the data
@@ -242,25 +258,29 @@ def load_tokenized_prepared_datasets(
                 suffix = ""
                 if ":load_" in d.type:
                     suffix = f" Did you mean {d.type.replace(':load_', '.load_')}?"
-                logging.error(
+                LOG.error(f"unhandled prompt tokenization strategy: {d.type}. {suffix}")
-                    f"unhandled prompt tokenization strategy: {d.type}. {suffix}"
-                )
                 raise ValueError(
                     f"unhandled prompt tokenization strategy: {d.type} {suffix}"
                 )
-        logging.info("tokenizing, merging, and shuffling master dataset")
+        LOG.info("tokenizing, merging, and shuffling master dataset")
 
         samples: List[int] = []
+        chunk_size = 1000
         for d in datasets:
-            samples = samples + list(d)
+            d_iter = iter(d)
+            while True:
+                chunk = list(itertools.islice(d_iter, chunk_size))
+                if not chunk:
+                    break
+                samples.extend(chunk)
+
+        LOG.info("shuffle")
         dataset = Dataset.from_list(samples).shuffle(seed=seed)
         if cfg.local_rank == 0:
-            logging.info(
+            LOG.info(f"Saving merged prepared dataset to disk... {prepared_ds_path}")
-                f"Saving merged prepared dataset to disk... {prepared_ds_path}"
-            )
             dataset.save_to_disk(prepared_ds_path)
             if cfg.push_dataset_to_hub:
-                logging.info(
+                LOG.info(
                     f"Saving merged prepared dataset with push_to_hub... {cfg.push_dataset_to_hub}/{ds_hash}"
                 )
                 dataset.push_to_hub(
@@ -311,7 +331,7 @@ def load_prepare_datasets(
         use_auth_token = cfg.hf_use_auth_token
         try:
             if cfg.push_dataset_to_hub:
-                logging.info(
+                LOG.info(
                     f"Checking for packed prepared dataset from hub... {cfg.push_dataset_to_hub}/{ds_hash}"
                 )
                 dataset = load_dataset(
@@ -325,13 +345,13 @@ def load_prepare_datasets(
         if dataset:
             ...
         elif any(prepared_ds_path.glob("*")):
-            logging.info(
+            LOG.info(
                 f"Loading prepared packed dataset from disk at {prepared_ds_path}..."
             )
             dataset = load_from_disk(str(prepared_ds_path))
-            logging.info("Prepared packed dataset loaded from disk...")
+            LOG.info("Prepared packed dataset loaded from disk...")
             if cfg.push_dataset_to_hub:
-                logging.info(
+                LOG.info(
                     f"Saving packed prepared dataset with push_to_hub... {cfg.push_dataset_to_hub}/{ds_hash}"
                 )
                 dataset.push_to_hub(
@@ -350,9 +370,7 @@ def load_prepare_datasets(
                 [dataset],
                 seq_length=max_packed_sequence_len,
             )
-            logging.info(
+            LOG.info(f"packing master dataset to len: {cfg.max_packed_sequence_len}")
-                f"packing master dataset to len: {cfg.max_packed_sequence_len}"
-            )
             dataset = Dataset.from_list(list(constant_len_dataset))
 
             # filter out bad data
@@ -368,12 +386,12 @@ def load_prepare_datasets(
             )
 
             if cfg.local_rank == 0:
-                logging.info(
+                LOG.info(
                     f"Saving packed prepared dataset to disk... {prepared_ds_path}"
                 )
                 dataset.save_to_disk(prepared_ds_path)
                 if cfg.push_dataset_to_hub:
-                    logging.info(
+                    LOG.info(
                         f"Saving packed prepared dataset with push_to_hub... {cfg.push_dataset_to_hub}/{ds_hash}"
                     )
                     dataset.push_to_hub(
@@ -386,7 +404,7 @@ def load_prepare_datasets(
         )
 
     if cfg.dataset_shard_num and cfg.dataset_shard_idx is not None:
-        logging.info(
+        LOG.info(
             f"Using index #{cfg.dataset_shard_idx} of {cfg.dataset_shard_num} shards"
         )
         dataset = dataset.shard(
@@ -394,8 +412,127 @@ def load_prepare_datasets(
             index=cfg.dataset_shard_idx,
         )
 
-    dataset = dataset.train_test_split(test_size=cfg.val_set_size, shuffle=False)
+    if cfg.val_set_size:
-    train_dataset = dataset["train"]
+        dataset = dataset.train_test_split(test_size=cfg.val_set_size, shuffle=False)
-    eval_dataset = dataset["test"]
+        train_dataset = dataset["train"]
+        eval_dataset = dataset["test"]
+    else:
+        train_dataset = dataset
+        eval_dataset = None
 
     return train_dataset, eval_dataset
+
+
+def encode_pretraining(tokenizer, max_tokens, examples):
+    res = tokenizer(
+        examples["text"],
+        truncation=True,
+        max_length=max_tokens - 2,
+        add_special_tokens=True,
+    )
+    # Convert to PyTorch tensors
+    input_ids = [torch.tensor(seq) for seq in res["input_ids"]]
+    attention_mask = [torch.tensor(seq) for seq in res["attention_mask"]]
+    new_input_ids = []
+    new_attention_mask = []
+    # Append EOS and PAD tokens to input_ids, and correct attention_mask
+    for i, _ in enumerate(input_ids):
+        input_ids[i] = torch.cat(
+            (
+                input_ids[i],
+                torch.tensor([tokenizer.eos_token_id, tokenizer.pad_token_id]),
+            ),
+            dim=0,
+        )
+        attention_mask[i] = torch.cat((attention_mask[i], torch.tensor([1, 0])), dim=0)
+
+    # Concatenate tokens so that their lengths are less than max_tokens
+    buffer_input_ids = torch.tensor([], dtype=torch.long)
+    buffer_attention_mask = torch.tensor([], dtype=torch.long)
+
+    for ids, mask in zip(input_ids, attention_mask):
+        if buffer_input_ids.numel() == max_tokens:
+            new_input_ids.append(buffer_input_ids)
+            new_attention_mask.append(buffer_attention_mask)
+            buffer_input_ids = torch.tensor([], dtype=torch.long)
+            buffer_attention_mask = torch.tensor([], dtype=torch.long)
+            buffer_input_ids = torch.cat((buffer_input_ids, ids), dim=0)
+            buffer_attention_mask = torch.cat((buffer_attention_mask, mask), dim=0)
+        elif buffer_input_ids.numel() + ids.numel() <= max_tokens:
+            buffer_input_ids = torch.cat((buffer_input_ids, ids), dim=0)
+            buffer_attention_mask = torch.cat((buffer_attention_mask, mask), dim=0)
+        else:
+            buffer_input_ids = torch.cat(
+                (
+                    buffer_input_ids,
+                    torch.full(
+                        (max_tokens - buffer_input_ids.numel(),),
+                        tokenizer.pad_token_id,
+                        dtype=torch.long,
+                    ),
+                ),
+                dim=0,
+            )
+            buffer_attention_mask = torch.cat(
+                (
+                    buffer_attention_mask,
+                    torch.full(
+                        (max_tokens - buffer_attention_mask.numel(),),
+                        0,
+                        dtype=torch.long,
+                    ),
+                ),
+                dim=0,
+            )
+            new_input_ids.append(buffer_input_ids)
+            new_attention_mask.append(buffer_attention_mask)
+            buffer_input_ids = torch.tensor([], dtype=torch.long)
+            buffer_attention_mask = torch.tensor([], dtype=torch.long)
+
+            buffer_input_ids = torch.cat((buffer_input_ids, ids), dim=0)
+            buffer_attention_mask = torch.cat((buffer_attention_mask, mask), dim=0)
+
+    if buffer_input_ids.numel() > 0:  # for any leftover tokens
+        while buffer_input_ids.numel() < max_tokens:  # make all sequences equal in size
+            buffer_input_ids = torch.cat(
+                (
+                    buffer_input_ids,
+                    torch.full(
+                        (max_tokens - buffer_input_ids.numel(),),
+                        tokenizer.pad_token_id,
+                        dtype=torch.long,
+                    ),
+                ),
+                dim=0,
+            )
+            buffer_attention_mask = torch.cat(
+                (
+                    buffer_attention_mask,
+                    torch.full(
+                        (max_tokens - buffer_attention_mask.numel(),),
+                        0,
+                        dtype=torch.long,
+                    ),
+                ),
+                dim=0,
+            )
+        new_input_ids.append(buffer_input_ids)
+        new_attention_mask.append(buffer_attention_mask)
+
+    ret = {
+        "input_ids": [seq.tolist() for seq in new_input_ids],
+        "labels": [seq.tolist() for seq in new_input_ids],
+        "attention_mask": [seq.tolist() for seq in new_attention_mask],
+    }
+
+    LOG.debug(len(ret["input_ids"]))
+    return ret
+
+
+def load_pretraining_dataset(path, tokenizer, max_tokens=2048, seed=42):
+    encode = functools.partial(encode_pretraining, tokenizer, max_tokens)
+    dataset = load_dataset(path, streaming=True, split="train")
+    dataset = dataset.shuffle(seed=seed, buffer_size=10_000)
+    # TODO dynamically figure out which columns/features to remove
+    dataset = dataset.map(encode, batched=True, remove_columns=["text", "meta"])
+    return dataset

src/axolotl/utils/models.py

@@ -10,17 +10,21 @@ from typing import TYPE_CHECKING, Optional, Tuple  # noqa: F401
 import bitsandbytes as bnb
 import torch
 import transformers
-from transformers import PreTrainedModel  # noqa: F401
+from optimum.bettertransformer import BetterTransformer
 from transformers import (  # noqa: F401
     AutoConfig,
     AutoModelForCausalLM,
     AutoTokenizer,
     BitsAndBytesConfig,
     LlamaConfig,
+    PreTrainedModel,
+    PreTrainedTokenizerBase,
 )
 
 from axolotl.prompt_tokenizers import LLAMA_DEFAULT_PAD_TOKEN
 
+LOG = logging.getLogger("axolotl")
+
 if TYPE_CHECKING:
     from peft import PeftConfig  # noqa: F401
 
@@ -32,21 +36,26 @@ def load_tokenizer(
     tokenizer_type,
     cfg,
 ):
+    use_fast = True  # this is the default
+    if cfg.tokenizer_use_fast is not None:
+        use_fast = cfg.tokenizer_use_fast
     if tokenizer_type:
         tokenizer = getattr(transformers, tokenizer_type).from_pretrained(
             tokenizer_config,
             trust_remote_code=cfg.trust_remote_code or False,
+            use_fast=use_fast,
         )
     else:
         tokenizer = AutoTokenizer.from_pretrained(
             tokenizer_config,
             trust_remote_code=cfg.trust_remote_code or False,
+            use_fast=use_fast,
         )
 
-    logging.debug(f"EOS: {tokenizer.eos_token_id} / {tokenizer.eos_token}")
+    LOG.debug(f"EOS: {tokenizer.eos_token_id} / {tokenizer.eos_token}")
-    logging.debug(f"BOS: {tokenizer.bos_token_id} / {tokenizer.bos_token}")
+    LOG.debug(f"BOS: {tokenizer.bos_token_id} / {tokenizer.bos_token}")
-    logging.debug(f"PAD: {tokenizer.pad_token_id} / {tokenizer.pad_token}")
+    LOG.debug(f"PAD: {tokenizer.pad_token_id} / {tokenizer.pad_token}")
-    logging.debug(f"UNK: {tokenizer.unk_token_id} / {tokenizer.unk_token}")
+    LOG.debug(f"UNK: {tokenizer.unk_token_id} / {tokenizer.unk_token}")
 
     if tokenizer.__class__.__name__ in [
         "LlamaTokenizer",
@@ -70,7 +79,7 @@ def load_tokenizer(
 def load_model(
     base_model, base_model_config, model_type, tokenizer, cfg, adapter="lora"
 ):
-    # type: (str, str, str, AutoTokenizer, DictDefault, Optional[str]) -> Tuple[PreTrainedModel, Optional[PeftConfig]]
+    # type: (str, str, str, PreTrainedTokenizerBase, DictDefault, Optional[str]) -> Tuple[PreTrainedModel, Optional[PeftConfig]]
     """
     Load a model from a base model and a model type.
     """
@@ -83,23 +92,25 @@ def load_model(
 
     if cfg.is_llama_derived_model and cfg.flash_attention:
         if cfg.device not in ["mps", "cpu"] and not cfg.inference:
-            from axolotl.flash_attn import replace_llama_attn_with_flash_attn
+            from axolotl.monkeypatch.llama_attn_hijack_flash import (
+                replace_llama_attn_with_flash_attn,
+            )
 
-            logging.info("patching with flash attention")
+            LOG.info("patching with flash attention")
             replace_llama_attn_with_flash_attn()
     elif cfg.is_llama_derived_model and cfg.xformers_attention:
         from axolotl.monkeypatch.llama_attn_hijack_xformers import (
             hijack_llama_attention,
         )
 
-        logging.info("patching with xformers attention")
+        LOG.info("patching with xformers attention")
         hijack_llama_attention()
     elif cfg.is_llama_derived_model and cfg.sdp_attention:
         from axolotl.monkeypatch.llama_attn_hijack_xformers import (
             hijack_llama_sdp_attention,
         )
 
-        logging.info("patching with sdp attention")
+        LOG.info("patching with sdp attention")
         hijack_llama_sdp_attention()
     elif cfg.is_llama_derived_model and cfg.landmark_attention:
         from axolotl.monkeypatch.llama_landmark_attn import (
@@ -107,7 +118,7 @@ def load_model(
             patch_llama_with_landmark_attn,
         )
 
-        logging.info("patching with landmark attention")
+        LOG.info("patching with landmark attention")
         patch_llama_with_landmark_attn()
 
         # Note: This might overwrite previous additional_special_tokens
@@ -118,12 +129,12 @@ def load_model(
             replace_llama_rope_with_xpos_rope,
         )
 
-        logging.info("patching with xpos rope")
+        LOG.info("patching with xpos rope")
         replace_llama_rope_with_xpos_rope()
 
-    if cfg.bf16:
+    if cfg.bf16 or cfg.bfloat16:
         torch_dtype = torch.bfloat16
-    elif cfg.load_in_8bit or cfg.fp16:
+    elif cfg.load_in_8bit or cfg.fp16 or cfg.float16:
         torch_dtype = torch.float16
     else:
         torch_dtype = torch.float32
@@ -135,18 +146,24 @@ def load_model(
 
             replace_peft_model_with_int4_lora_model()
     except Exception as err:
-        logging.exception(err)
+        LOG.exception(err)
         raise err
 
-    try:
+    if not cfg.gptq and (
-        from peft import prepare_model_for_kbit_training
+        (cfg.adapter == "lora" and load_in_8bit)
-    except ImportError:
+        or (cfg.adapter == "qlora" and cfg.load_in_4bit)
-        # For backward compatibility
+    ):
-        from peft import (
+        try:
-            prepare_model_for_int8_training as prepare_model_for_kbit_training,
+            from peft import prepare_model_for_kbit_training
-        )
+        except ImportError:
+            # For backward compatibility
+            from peft import (
+                prepare_model_for_int8_training as prepare_model_for_kbit_training,
+            )
 
     model_kwargs = {}
+    if cfg.model_revision:
+        model_kwargs["revision"] = cfg.model_revision
     if cfg.adapter == "qlora" and cfg.load_in_4bit:
         model_kwargs["quantization_config"] = BitsAndBytesConfig(
             load_in_4bit=True,
@@ -178,7 +195,7 @@ def load_model(
                 if len(files) > 0:
                     model_path = str(files[0])
                 else:
-                    logging.warning(
+                    LOG.warning(
                         "unable to find a cached model file, this will likely fail..."
                     )
                     model_path = str(cache_model_path)
@@ -195,7 +212,7 @@ def load_model(
                 else True,
             )
             load_in_8bit = False
-        elif cfg.is_llama_derived_model:
+        elif cfg.is_llama_derived_model and not cfg.trust_remote_code:
             from transformers import LlamaForCausalLM
 
             config = LlamaConfig.from_pretrained(base_model_config)
@@ -234,7 +251,7 @@ def load_model(
         #         device=cfg.device,
         #     )
         #     model.train() # sets to train instead of eval mode
-        elif model_type:
+        elif model_type and not cfg.trust_remote_code:
             model = getattr(transformers, model_type).from_pretrained(
                 base_model,
                 load_in_8bit=cfg.load_in_8bit and cfg.adapter is not None,
@@ -251,15 +268,20 @@ def load_model(
             )
             # Shouldn't be a problem most of the time. will obviously error if the model doesn't support this
             # when training starts
-            if hasattr(config, "max_seq_len") and cfg.sequence_len > config.max_seq_len:
+            if (
+                hasattr(config, "max_seq_len")
+                and config.max_seq_len
+                and cfg.sequence_len > config.max_seq_len
+            ):
                 config.max_seq_len = cfg.sequence_len
-                logging.warning(f"increasing context length to {cfg.sequence_len}")
+                LOG.warning(f"increasing context length to {cfg.sequence_len}")
             elif (
                 hasattr(config, "max_sequence_length")
+                and config.max_sequence_length
                 and cfg.sequence_len > config.max_sequence_length
             ):
                 config.max_sequence_length = cfg.sequence_len
-                logging.warning(f"increasing context length to {cfg.sequence_len}")
+                LOG.warning(f"increasing context length to {cfg.sequence_len}")
             model = AutoModelForCausalLM.from_pretrained(
                 base_model,
                 config=config,
@@ -271,31 +293,56 @@ def load_model(
                 **model_kwargs,
             )
     except Exception as err:  # pylint: disable=broad-exception-caught
-        logging.error(
+        LOG.error(
             "Exception raised attempting to load model, retrying with AutoModelForCausalLM"
         )
-        logging.exception(err)
+        LOG.exception(err)
         model = AutoModelForCausalLM.from_pretrained(
             base_model,
             load_in_8bit=cfg.load_in_8bit and cfg.adapter is not None,
+            load_in_4bit=cfg.load_in_4bit and cfg.adapter is not None,
             torch_dtype=torch_dtype,
             device_map=cfg.device_map,
             trust_remote_code=cfg.trust_remote_code or False,
             **model_kwargs,
         )
 
-    embeddings_len = math.ceil(len(tokenizer) / 32) * 32
+    embeddings_len = (
+        math.ceil(len(tokenizer) / 32) * 32
+        if cfg.resize_token_embeddings_to_32x
+        else len(tokenizer)
+    )
     model.resize_token_embeddings(embeddings_len)
 
+    if (
+        hasattr(model.config, "max_position_embeddings")
+        and model.config.max_position_embeddings
+        and cfg.sequence_len >= model.config.max_position_embeddings
+    ):
+        LOG.warning(
+            f"increasing model.config.max_position_embeddings to {cfg.sequence_len}"
+        )
+        model.config.max_position_embeddings = cfg.sequence_len
+
     if not cfg.gptq and (
         (cfg.adapter == "lora" and load_in_8bit)
         or (cfg.adapter == "qlora" and cfg.load_in_4bit)
     ):
-        logging.info("converting PEFT model w/ prepare_model_for_kbit_training")
+        LOG.info("converting PEFT model w/ prepare_model_for_kbit_training")
         model = prepare_model_for_kbit_training(
             model, use_gradient_checkpointing=cfg.gradient_checkpointing
         )
 
+        # LlamaRMSNorm layers are in fp32 after kbit_training, so we need to
+        # convert them back to fp16/bf16 for flash-attn compatibility.
+        if cfg.flash_attention and cfg.is_llama_derived_model:
+            for name, module in model.named_modules():
+                if "norm" in name:
+                    module.to(torch_dtype)
+                if "lm_head" in name or "embed_tokens" in name:
+                    if hasattr(module, "weight"):
+                        module.to(torch_dtype)
+
     model, lora_config = load_adapter(model, cfg, adapter)
 
     if cfg.ddp and not load_in_8bit:
@@ -303,7 +350,7 @@ def load_model(
 
     if cfg.gptq:
         # Scales to half
-        logging.info("Fitting 4bit scales and zeros to half")
+        LOG.info("Fitting 4bit scales and zeros to half")
         for _, module in model.named_modules():
             if "Autograd4bitQuantLinear" in str(type(module)) or "Linear4bitLt" in str(
                 type(module)
@@ -329,9 +376,12 @@ def load_model(
         if param.requires_grad:
             requires_grad.append(f"{name}: {param.requires_grad}")
     if len(requires_grad) == 0:
-        logging.warning("there are no parameters that require gradient updates")
+        LOG.warning("there are no parameters that require gradient updates")
     model.config.use_cache = False
 
+    if cfg.flash_optimum:
+        model = BetterTransformer.transform(model)
+
     # TODO resume_from_checkpoint handling
     return model, lora_config
 
@@ -360,7 +410,7 @@ def load_llama_adapter(model, cfg):
     )
 
     if cfg.lora_model_dir:
-        logging.info("Loading pretained LORA")
+        LOG.info("Loading pretained LORA")
         model = PeftModel.from_pretrained(
             model,
             cfg.lora_model_dir,
@@ -407,7 +457,7 @@ def load_lora(model, cfg):
             bits = 8
 
         linear_names = find_all_linear_names(bits, model)
-        logging.info(f"found linear modules: {repr(linear_names)}")
+        LOG.info(f"found linear modules: {repr(linear_names)}")
         lora_target_modules = list(set(lora_target_modules + linear_names))
 
     lora_config = LoraConfig(

src/axolotl/utils/schedulers.py

@@ -1,6 +1,9 @@
 """Module for custom LRScheduler class"""
+import math
+from functools import partial
 
-from torch.optim.lr_scheduler import LRScheduler
+from torch.optim import Optimizer
+from torch.optim.lr_scheduler import LambdaLR, LRScheduler
 
 
 class InterpolatingLogScheduler(LRScheduler):
@@ -42,3 +45,58 @@ class InterpolatingLogScheduler(LRScheduler):
             lrs = [self.max_lr for base_lr in self.base_lrs]
 
         return lrs
+
+
+def _get_cosine_schedule_with_quadratic_warmup_lr_lambda(
+    current_step: int,
+    *,
+    num_warmup_steps: int,
+    num_training_steps: int,
+    num_cycles: float
+):
+    if current_step < num_warmup_steps:
+        return (float(current_step) / float(max(1, num_warmup_steps))) ** 2
+    progress = float(current_step - num_warmup_steps) / float(
+        max(1, num_training_steps - num_warmup_steps)
+    )
+    return max(
+        0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress))
+    )
+
+
+def get_cosine_schedule_with_quadratic_warmup(
+    optimizer: Optimizer,
+    num_warmup_steps: int,
+    num_training_steps: int,
+    num_cycles: float = 0.5,
+    last_epoch: int = -1,
+):
+    """
+    Create a schedule with a learning rate that decreases following the values of the cosine function between the
+    initial lr set in the optimizer to 0, after a warmup period during which it increases linearly between 0 and the
+    initial lr set in the optimizer.
+
+    Args:
+        optimizer ([`~torch.optim.Optimizer`]):
+            The optimizer for which to schedule the learning rate.
+        num_warmup_steps (`int`):
+            The number of steps for the warmup phase.
+        num_training_steps (`int`):
+            The total number of training steps.
+        num_cycles (`float`, *optional*, defaults to 0.5):
+            The number of waves in the cosine schedule (the defaults is to just decrease from the max value to 0
+            following a half-cosine).
+        last_epoch (`int`, *optional*, defaults to -1):
+            The index of the last epoch when resuming training.
+
+    Return:
+        `torch.optim.lr_scheduler.LambdaLR` with the appropriate schedule.
+    """
+
+    lr_lambda = partial(
+        _get_cosine_schedule_with_quadratic_warmup_lr_lambda,
+        num_warmup_steps=num_warmup_steps,
+        num_training_steps=num_training_steps,
+        num_cycles=num_cycles,
+    )
+    return LambdaLR(optimizer, lr_lambda, last_epoch)

src/axolotl/utils/tokenization.py

@@ -5,6 +5,8 @@ import logging
 
 from termcolor import colored
 
+LOG = logging.getLogger("axolotl")
+
 
 def check_dataset_labels(dataset, tokenizer):
     # the dataset is already shuffled, so let's just check the first 5 elements
@@ -32,5 +34,7 @@ def check_example_labels(example, tokenizer):
         )
         colored_tokens.append(colored_token)
 
-    logging.info(" ".join(colored_tokens))
+    LOG.info(" ".join(colored_tokens))
-    logging.info("\n\n\n")
+    LOG.info("\n\n\n")
+
+    return " ".join(colored_tokens)

src/axolotl/utils/trainer.py

@@ -5,6 +5,7 @@ import logging
 import math
 import os
 import sys
+from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Optional
 
@@ -13,14 +14,70 @@ import torch.cuda
 import transformers
 from torch import nn
 from torch.optim.lr_scheduler import OneCycleLR
-from transformers import EarlyStoppingCallback, Trainer
+from transformers import EarlyStoppingCallback, Trainer, TrainingArguments
 from transformers.trainer_pt_utils import get_parameter_names
 
-from axolotl.utils.callbacks import SavePeftModelCallback
+from axolotl.utils.callbacks import (
-from axolotl.utils.schedulers import InterpolatingLogScheduler
+    SaveBetterTransformerModelCallback,
+    SavePeftModelCallback,
+)
+from axolotl.utils.schedulers import (
+    InterpolatingLogScheduler,
+    get_cosine_schedule_with_quadratic_warmup,
+)
+
+LOG = logging.getLogger("axolotl")
 
 
-class OneCycleLRSchedulerTrainer(Trainer):
+@dataclass
+class AxolotlTrainingArguments(TrainingArguments):
+    """
+    Extend the base TrainingArguments for axolotl helpers
+    """
+
+    lr_quadratic_warmup: bool = field(
+        default=False,
+        metadata={"help": "Use quadratic warmup for cosine scheduling."},
+    )
+
+
+class AxolotlTrainer(Trainer):
+    """
+    Extend the base Trainer for axolotl helpers
+    """
+
+    args = None  # type: AxolotlTrainingArguments
+
+    def create_scheduler(
+        self, num_training_steps: int, optimizer: torch.optim.Optimizer = None
+    ):
+        """
+        Setup the scheduler. The optimizer of the trainer must have been set up either before this method is called or
+        passed as an argument.
+
+        Args:
+            num_training_steps (int): The number of training steps to do.
+            optimizer (torch.optim.Optimizer): The training optimizer
+        """
+
+        # fmt: off
+        if self.lr_scheduler is None:  # type: ignore  # pylint: disable=access-member-before-definition
+            # fmt: on
+            if (
+                self.args.lr_scheduler_type == "cosine"
+                and self.args.lr_quadratic_warmup is True
+            ):
+                self.lr_scheduler = get_cosine_schedule_with_quadratic_warmup(  # pylint: disable=attribute-defined-outside-init
+                    optimizer,
+                    num_warmup_steps=self.args.get_warmup_steps(num_training_steps),
+                    num_training_steps=num_training_steps,
+                )
+            else:
+                return super().create_scheduler(num_training_steps, optimizer)
+        return self.lr_scheduler
+
+
+class OneCycleLRSchedulerTrainer(AxolotlTrainer):
     """
     Trainer subclass that uses the OneCycleLR scheduler
     """
@@ -100,6 +157,9 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
         if cfg.fsdp_config:
             training_arguments_kwargs["fsdp_config"] = dict(cfg.fsdp_config)
 
+    if cfg.lr_quadratic_warmup is not None:
+        training_arguments_kwargs["lr_quadratic_warmup"] = cfg.lr_quadratic_warmup
+
     # deepspeed
     if (
         os.environ.get("ACCELERATE_USE_DEEPSPEED") == "true"
@@ -112,7 +172,24 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
             # TODO search Path("./") for one
             training_arguments_kwargs["deepspeed"] = "./ds_config.json"
 
-    training_args = transformers.TrainingArguments(
+    if cfg.adam_beta1:
+        training_arguments_kwargs["adam_beta1"] = cfg.adam_beta1
+    if cfg.adam_beta2:
+        training_arguments_kwargs["adam_beta2"] = cfg.adam_beta2
+    if cfg.adam_epsilon:
+        training_arguments_kwargs["adam_epsilon"] = cfg.adam_epsilon
+    if cfg.max_grad_norm:
+        training_arguments_kwargs["max_grad_norm"] = cfg.max_grad_norm
+
+    if cfg.hub_model_id:
+        training_arguments_kwargs["hub_model_id"] = cfg.hub_model_id
+        training_arguments_kwargs["push_to_hub"] = True
+        training_arguments_kwargs["hub_private_repo"] = True
+
+    if cfg.save_safetensors:
+        training_arguments_kwargs["save_safetensors"] = cfg.save_safetensors
+
+    training_args = AxolotlTrainingArguments(  # pylint: disable=unexpected-keyword-arg
         per_device_train_batch_size=cfg.micro_batch_size,
         per_device_eval_batch_size=cfg.eval_batch_size
         if cfg.eval_batch_size is not None
@@ -228,6 +305,9 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
     ]:  # only save in rank 0
         callbacks.append(SavePeftModelCallback)
 
+    if hasattr(model, "use_bettertransformer") and model.use_bettertransformer is True:
+        callbacks.append(SaveBetterTransformerModelCallback)
+
     data_collator_kwargs = {
         "padding": True,
     }
@@ -247,7 +327,7 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
 
         set_model_mem_id(model, tokenizer)
 
-        logging.info("Adding landmark attention tokens to dataset")
+        LOG.info("Adding landmark attention tokens to dataset")
 
         for dataset in [train_dataset, eval_dataset]:
             dataset = dataset.map(
@@ -259,7 +339,7 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
     trainer_cls = (
         OneCycleLRSchedulerTrainer
         if cfg.lr_scheduler == "one_cycle" and (cfg.fsdp or cfg.adapter == "qlora")
-        else transformers.Trainer
+        else AxolotlTrainer
     )
     trainer = trainer_cls(
         model=model,

src/axolotl/utils/validation.py

@@ -2,6 +2,10 @@
 
 import logging
 
+import torch
+
+LOG = logging.getLogger("axolotl")
+
 
 def validate_config(cfg):
     if cfg.gradient_accumulation_steps and cfg.batch_size:
@@ -9,7 +13,7 @@ def validate_config(cfg):
             "please set only one of gradient_accumulation_steps or batch_size"
         )
     if cfg.batch_size:
-        logging.warning(
+        LOG.warning(
             "%s\n%s",
             "batch_size is not recommended. Please use gradient_accumulation_steps instead.",
             "To calculate the equivalent gradient_accumulation_steps, divide batch_size / micro_batch_size / number of gpus.",
@@ -42,10 +46,10 @@ def validate_config(cfg):
                 raise ValueError("Require cfg.load_in_4bit to be True for qlora")
 
     if not cfg.load_in_8bit and cfg.adapter == "lora":
-        logging.warning("We recommend setting `load_in_8bit: true` for LORA finetuning")
+        LOG.warning("We recommend setting `load_in_8bit: true` for LORA finetuning")
 
     if cfg.trust_remote_code:
-        logging.warning(
+        LOG.warning(
             "`trust_remote_code` is set to true. Please make sure that you reviewed the remote code/model."
         )
 
@@ -62,7 +66,45 @@ def validate_config(cfg):
     ) and cfg.gradient_checkpointing:
         raise ValueError("gradient_checkpointing is not supported for MPT models")
 
+    if cfg.flash_optimum is True:
+        if cfg.adapter:
+            LOG.warning("BetterTransformers probably doesn't work with PEFT adapters")
+        if cfg.fp16 or cfg.bf16:
+            raise ValueError("AMP is not supported with BetterTransformer")
+        if cfg.float16 is not True and cfg.bloat16 is not True:
+            LOG.warning(
+                "You should probably set bfloat16 or float16 to true to "
+                "load the model in float16 for BetterTransformers"
+            )
+        if int(torch.__version__.split(".")[0]) < 2:
+            LOG.warning("torch>=2.0.0 required")
+            raise ValueError(
+                f"flash_optimum for BetterTransformers may not be used with {torch.__version__}"
+            )
+
+    if cfg.pretraining_dataset and cfg.group_by_length:
+        LOG.warning(
+            "You probably want to disable group_by_length as it will force a streamed dataset to download completely."
+        )
+
+    if any([cfg.adam_beta1, cfg.adam_beta2, cfg.adam_epsilon]) and (
+        not cfg.optimizer or "adamw" not in cfg.optimizer
+    ):
+        LOG.warning("adamw hyperparameters found, but no adamw optimizer set")
+
+    if cfg.push_to_hub_model_id:
+        raise ValueError(
+            "push_to_hub_model_id is deprecated. Please use hub_model_id instead."
+        )
+
     # TODO
     # MPT 7b
     # https://github.com/facebookresearch/bitsandbytes/issues/25
-    # no 8bit adamw w bf16
+    # no 8bit adaAmw w bf16
+
+    # GPT-NeoX
+    # evals broken when extending context len
+    # File "/root/miniconda3/envs/py3.9/lib/python3.9/site-packages/transformers/models/gpt_neox/modeling_gpt_neox.py", line 162, in forward                        attn_output, attn_weights = self._attn(query, key, value, attention_mask, head_mask)
+    # File "/root/miniconda3/envs/py3.9/lib/python3.9/site-packages/optimum/bettertransformer/models/attention.py", line 74, in gpt2_wrapped_scaled_dot_product
+    # attention_mask = causal_mask + attention_mask
+    # RuntimeError: The size of tensor a (2048) must match the size of tensor b (8132) at non-singleton dimension 3

tests/test_prompt_tokenizers.py

@@ -6,10 +6,18 @@ from pathlib import Path
 
 from transformers import AutoTokenizer
 
-from axolotl.prompt_tokenizers import ShareGPTPromptTokenizingStrategy
+from axolotl.prompt_strategies.alpaca_chat import NoSystemPrompter
-from axolotl.prompters import ShareGPTPrompter
+from axolotl.prompt_strategies.alpaca_w_system import (
+    InstructionWSystemPromptTokenizingStrategy,
+    SystemDataPrompter,
+)
+from axolotl.prompt_tokenizers import (
+    AlpacaPromptTokenizingStrategy,
+    ShareGPTPromptTokenizingStrategy,
+)
+from axolotl.prompters import AlpacaPrompter, PromptStyle, ShareGPTPrompter
 
-logging.basicConfig(level="INFO")
+LOG = logging.getLogger("axolotl")
 
 
 class TestPromptTokenizationStrategies(unittest.TestCase):
@@ -29,7 +37,6 @@ class TestPromptTokenizationStrategies(unittest.TestCase):
         )
 
     def test_sharegpt_integration(self):
-        print(Path(__file__).parent)
         with open(
             Path(__file__).parent / "fixtures/conversation.json", encoding="utf-8"
         ) as fin:
@@ -53,6 +60,80 @@ class TestPromptTokenizationStrategies(unittest.TestCase):
             self.assertEqual(len(example[fields]), len(tokenized_conversation[fields]))
             self.assertEqual(example[fields], tokenized_conversation[fields])
 
+    def test_no_sys_prompt(self):
+        """
+        tests the interface between the user and assistant parts
+        """
+        prompter = NoSystemPrompter()
+        # pylint: disable=duplicate-code
+        strat = AlpacaPromptTokenizingStrategy(
+            prompter,
+            self.tokenizer,
+            False,
+            2048,
+        )
+        sample = {
+            "instruction": "hello cruel. lorem ipsum dolor sit amet.",
+            "output": "world!",
+        }
+        example = strat.tokenize_prompt(sample)
+        world_idx = example["input_ids"].index(3186)
+        assert example["labels"][world_idx] == 3186
+        assert example["labels"][world_idx - 1] == -100
+
+    def test_alpaca(self):
+        """
+        tests the interface between the user and assistant parts
+        """
+        # pylint: disable=duplicate-code
+        prompter = AlpacaPrompter()
+        strat = AlpacaPromptTokenizingStrategy(
+            prompter,
+            self.tokenizer,
+            False,
+            2048,
+        )
+        sample = {"instruction": "hello!", "output": "Hi! How can I help?"}
+        example = strat.tokenize_prompt(sample)
+        world_idx = example["input_ids"].index(6324)
+        assert example["labels"][world_idx] == 6324
+        assert example["labels"][world_idx - 1] == -100
+
+
+class InstructionWSystemPromptTokenizingStrategyTest(unittest.TestCase):
+    """
+    Test class for prompt tokenization strategies with sys prompt from the dataset
+    """
+
+    def setUp(self) -> None:
+        # pylint: disable=duplicate-code
+        self.tokenizer = AutoTokenizer.from_pretrained("huggyllama/llama-7b")
+        self.tokenizer.add_special_tokens(
+            {
+                "bos_token": "<s>",
+                "eos_token": "</s>",
+                "unk_token": "<unk>",
+            }
+        )
+
+    def test_system_alpaca(self):
+        prompter = SystemDataPrompter(PromptStyle.CHAT.value)
+        strat = InstructionWSystemPromptTokenizingStrategy(
+            prompter,
+            self.tokenizer,
+            False,
+            2048,
+        )
+        sample = {
+            "system": "use cot",
+            "instruction": "hello!",
+            "output": "Hi! How can I help?",
+        }
+        example = strat.tokenize_prompt(sample)
+        assert example["input_ids"][0:4] == [1, 835, 2184, 29901]  # "<s>### System:"
+        assert example["input_ids"][5:7] == [1509, 20118]  # "use cot"
+        assert example["input_ids"][9] == 11889  # USER
+
 
 if __name__ == "__main__":
     unittest.main()

tests/test_prompters.py

@@ -2,7 +2,13 @@
 
 import unittest
 
-from axolotl.prompters import AlpacaPrompter, PromptStyle
+from axolotl.prompt_strategies.alpaca_w_system import SystemDataPrompter
+from axolotl.prompters import (
+    AlpacaPrompter,
+    MultipleChoiceExplainPrompter,
+    PromptStyle,
+    UnpromptedPrompter,
+)
 
 
 class AlpacaPrompterTest(unittest.TestCase):
@@ -55,3 +61,64 @@ class AlpacaPrompterTest(unittest.TestCase):
         assert "### Response:" not in res
         assert "USER:" in res
         assert "ASSISTANT:" in res
+
+    def test_system_prompt(self):
+        prompter = SystemDataPrompter(prompt_style=PromptStyle.CHAT.value)
+        res = next(
+            prompter.build_prompt_w_system(
+                "use cot", "tell me a joke about the following", "alpacas"
+            )
+        )
+        assert "use cot" in res
+        assert res.startswith("### System:")
+        assert "### Instruction:" not in res
+        assert "### Input:" not in res
+        assert "alpacas" in res
+        assert "### Response:" not in res
+        assert "USER:" in res
+        assert "ASSISTANT:" in res
+
+
+class UnpromptedPrompterTest(unittest.TestCase):
+    """
+    Test class for UnpromptedPrompter with no system prompts
+    """
+
+    def test_prompt_style_w_none(self):
+        prompter = UnpromptedPrompter(prompt_style=None)
+        res = next(prompter.build_prompt("tell me a joke"))
+        assert "### Instruction:" in res
+        assert "tell me a joke" in res
+        assert res.startswith("###")
+
+    def test_prompt_style_w_instruct(self):
+        prompter = UnpromptedPrompter(prompt_style=PromptStyle.INSTRUCT.value)
+        res = next(
+            prompter.build_prompt("tell me a joke about the following", "alpacas")
+        )
+        assert "### Instruction:" in res
+        assert "tell me a joke" in res
+        assert res.startswith("###")
+
+    def test_prompt_style_w_chat(self):
+        prompter = UnpromptedPrompter(prompt_style=PromptStyle.CHAT.value)
+        res = next(
+            prompter.build_prompt("tell me a joke about the following", "alpacas")
+        )
+        assert "USER:" in res
+        assert "tell me a joke" in res
+        assert res.startswith("USER:")
+
+
+class MultipleChoiceExplainPrompterTest(unittest.TestCase):
+    """
+    Test class for MultipleChoiceExplainPrompter
+    """
+
+    def test_prompt_style_w_chat(self):
+        prompter = MultipleChoiceExplainPrompter(prompt_style=PromptStyle.CHAT.value)
+        res = next(prompter.build_prompt("choose one", "- A\n- B\n- C", "C"))
+        assert "USER:" in res
+        assert "choose one" in res
+        assert "Choose the answer that best answers the question." in res
+        assert "- A\n- B\n- C" in res

tests/test_tokenizers.py

@@ -0,0 +1,31 @@
+"""
+Test cases for the tokenizer loading
+"""
+import unittest
+
+from axolotl.utils.dict import DictDefault
+from axolotl.utils.models import load_tokenizer
+
+
+class TestTokenizers(unittest.TestCase):
+    """
+    test class for the load_tokenizer fn
+    """
+
+    def test_default_use_fast(self):
+        cfg = DictDefault({})
+        tokenizer = load_tokenizer("huggyllama/llama-7b", None, cfg)
+        assert "Fast" in tokenizer.__class__.__name__
+
+    def test_dont_use_fast(self):
+        cfg = DictDefault(
+            {
+                "tokenizer_use_fast": False,
+            }
+        )
+        tokenizer = load_tokenizer("huggyllama/llama-7b", None, cfg)
+        assert "Fast" not in tokenizer.__class__.__name__
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_validation.py

@@ -212,3 +212,104 @@ class ValidationTest(unittest.TestCase):
 
         with pytest.raises(ValueError, match=regex_exp):
             validate_config(cfg)
+
+    def test_flash_optimum(self):
+        cfg = DictDefault(
+            {
+                "flash_optimum": True,
+                "adapter": "lora",
+            }
+        )
+
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "BetterTransformers probably doesn't work with PEFT adapters"
+                in record.message
+                for record in self._caplog.records
+            )
+
+        cfg = DictDefault(
+            {
+                "flash_optimum": True,
+            }
+        )
+
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "probably set bfloat16 or float16" in record.message
+                for record in self._caplog.records
+            )
+
+        cfg = DictDefault(
+            {
+                "flash_optimum": True,
+                "fp16": True,
+            }
+        )
+        regex_exp = r".*AMP is not supported.*"
+
+        with pytest.raises(ValueError, match=regex_exp):
+            validate_config(cfg)
+
+        cfg = DictDefault(
+            {
+                "flash_optimum": True,
+                "bf16": True,
+            }
+        )
+        regex_exp = r".*AMP is not supported.*"
+
+        with pytest.raises(ValueError, match=regex_exp):
+            validate_config(cfg)
+
+    def test_adamw_hyperparams(self):
+        cfg = DictDefault(
+            {
+                "optimizer": None,
+                "adam_epsilon": 0.0001,
+            }
+        )
+
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "adamw hyperparameters found, but no adamw optimizer set"
+                in record.message
+                for record in self._caplog.records
+            )
+
+        cfg = DictDefault(
+            {
+                "optimizer": "adafactor",
+                "adam_beta1": 0.0001,
+            }
+        )
+
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "adamw hyperparameters found, but no adamw optimizer set"
+                in record.message
+                for record in self._caplog.records
+            )
+
+        cfg = DictDefault(
+            {
+                "optimizer": "adamw_bnb_8bit",
+                "adam_beta1": 0.9,
+                "adam_beta2": 0.99,
+                "adam_epsilon": 0.0001,
+            }
+        )
+
+        validate_config(cfg)
+
+        cfg = DictDefault(
+            {
+                "optimizer": "adafactor",
+            }
+        )
+
+        validate_config(cfg)