Merge branch 'main' into print_venv

.bandit

@@ -1,3 +1,3 @@
 [bandit]
 exclude = tests
-skips = B101,B615
+skips = B101

.github/workflows/base.yml

@@ -29,11 +29,11 @@ jobs:
             cuda_version: 12.4.1
             cudnn_version: ""
             python_version: "3.11"
-            pytorch: 2.6.0
+            pytorch: 2.5.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
             dockerfile: "Dockerfile-base"
-          - cuda: "126"
-            cuda_version: 12.6.3
+          - cuda: "124"
+            cuda_version: 12.4.1
             cudnn_version: ""
             python_version: "3.11"
             pytorch: 2.6.0
@@ -43,7 +43,7 @@ jobs:
             cuda_version: 12.6.3
             cudnn_version: ""
             python_version: "3.11"
-            pytorch: 2.7.0
+            pytorch: 2.6.0
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
             dockerfile: "Dockerfile-base"
           - cuda: "126"

.github/workflows/main.yml

@@ -15,15 +15,15 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 126
-            cuda_version: 12.6.3
+          - cuda: 124
+            cuda_version: 12.4.1
             python_version: "3.11"
-            pytorch: 2.6.0
+            pytorch: 2.5.1
             axolotl_extras:
           - cuda: 126
             cuda_version: 12.6.3
             python_version: "3.11"
-            pytorch: 2.7.0
+            pytorch: 2.6.0
             axolotl_extras: vllm
           - cuda: 126
             cuda_version: 12.6.3
@@ -82,17 +82,17 @@ jobs:
     strategy:
       matrix:
         include:
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.5.1
+            axolotl_extras:
           - cuda: 126
             cuda_version: 12.6.3
             python_version: "3.11"
             pytorch: 2.6.0
             axolotl_extras:
             is_latest: true
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.0
-            axolotl_extras:
           - cuda: 126
             cuda_version: 12.6.3
             python_version: "3.11"

.github/workflows/multi-gpu-e2e.yml

@@ -33,11 +33,11 @@ jobs:
             axolotl_extras:
             num_gpus: 2
             nightly_build: "true"
-          - cuda: 126
-            cuda_version: 12.6.3
+          - cuda: 124
+            cuda_version: 12.4.1
             python_version: "3.11"
-            pytorch: 2.7.0
-            axolotl_extras: vllm
+            pytorch: 2.5.1
+            axolotl_extras:
             num_gpus: 2
             nightly_build: "true"
           - cuda: 126

.github/workflows/nightlies.yml

@@ -12,16 +12,16 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 126
-            cuda_version: 12.6.3
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.5.1
+            axolotl_extras:
+          - cuda: 124
+            cuda_version: 12.4.1
             python_version: "3.11"
             pytorch: 2.6.0
             axolotl_extras:
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.1
-            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
@@ -65,16 +65,16 @@ jobs:
     strategy:
       matrix:
         include:
-          - cuda: 126
-            cuda_version: 12.6.3
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.5.1
+            axolotl_extras:
+          - cuda: 124
+            cuda_version: 12.4.1
             python_version: "3.11"
             pytorch: 2.6.0
             axolotl_extras:
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.1
-            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout

.github/workflows/preview-docs.yml

@@ -28,8 +28,6 @@ jobs:
     steps:
       - name: Check out repository
         uses: actions/checkout@v4
-        with:
-          ref: ${{ github.event.pull_request.head.sha }}
 
       - name: Set up Quarto
         uses: quarto-dev/quarto-actions/setup@v2
@@ -52,11 +50,10 @@ jobs:
 
       - name: Netlify Publish
         uses: nwtgck/actions-netlify@v3.0
-        id: netlify
         with:
           publish-dir: './_site'
-          enable-pull-request-comment: false
-          enable-github-deployment: false
+          enable-pull-request-comment: true
+          enable-github-deployment: true
           github-token: ${{ secrets.GITHUB_TOKEN }}
           deploy-message: "Deployed On Netlify"
           github-deployment-environment: 'preview'
@@ -64,13 +61,3 @@ jobs:
         env:
           NETLIFY_AUTH_TOKEN: ${{ secrets.NETLIFY_AUTH_TOKEN }}
           NETLIFY_SITE_ID: ${{ secrets.NETLIFY_SITE_ID }}
-
-      - name: Update PR with preview link
-        if: ${{ steps.netlify.outcome == 'success' }}
-        uses: marocchino/sticky-pull-request-comment@v2
-        with:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          message: |
-            📖 **Documentation Preview**: ${{ steps.netlify.outputs.deploy-url }}
-
-            Deployed on Netlify from commit ${{ github.event.pull_request.head.sha }}

.github/workflows/tests-nightly.yml

@@ -26,7 +26,7 @@ jobs:
       max-parallel: 2
       matrix:
         python_version: ["3.11"]
-        pytorch_version: ["2.6.0", "2.7.0"]
+        pytorch_version: ["2.5.1", "2.6.0", "2.7.0"]
     timeout-minutes: 20
 
     steps:
@@ -80,9 +80,9 @@ jobs:
 
       - name: Run tests
         run: |
-          pytest -v --durations=10 -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
-          pytest -v --durations=10 tests/patched/
-          pytest -v --durations=10 tests/cli/
+          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
+          pytest -v tests/patched/
+          pytest -v tests/cli/
 
       - name: cleanup pip cache
         run: |
@@ -92,7 +92,7 @@ jobs:
     if: github.repository_owner == 'axolotl-ai-cloud'
     # this job needs to be run on self-hosted GPU runners...
     runs-on: [self-hosted, modal]
-    timeout-minutes: 120
+    timeout-minutes: 60
     needs: [pre-commit, pytest]
 
     strategy:
@@ -116,7 +116,7 @@ jobs:
       - name: Install Modal
         run: |
           python -m pip install --upgrade pip
-          pip install modal==1.0.2 jinja2
+          pip install modal==0.71.8 jinja2
       - name: Update env vars
         run: |
           echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV

.github/workflows/tests.yml

@@ -52,7 +52,7 @@ jobs:
       fail-fast: false
       matrix:
         python_version: ["3.11"]
-        pytorch_version: ["2.6.0", "2.7.0", "2.7.1"]
+        pytorch_version: ["2.5.1", "2.6.0", "2.7.1"]
     timeout-minutes: 20
 
     steps:
@@ -102,9 +102,9 @@ jobs:
 
       - name: Run tests
         run: |
-          pytest -v --durations=10 -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/ --cov=axolotl --cov-report=xml
-          pytest -v --durations=10 tests/patched/ --cov=axolotl --cov-append --cov-report=xml
-          pytest -v --durations=10 tests/cli/ --cov=axolotl --cov-append --cov-report=xml
+          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/ --cov=axolotl --cov-report=xml
+          pytest -v tests/patched/ --cov=axolotl --cov-append --cov-report=xml
+          pytest -v tests/cli/ --cov=axolotl --cov-append --cov-report=xml
 
       - name: Upload coverage to Codecov
         uses: codecov/codecov-action@v5
@@ -125,7 +125,7 @@ jobs:
       fail-fast: false
       matrix:
         python_version: ["3.11"]
-        pytorch_version: ["2.6.0", "2.7.0", "2.7.1"]
+        pytorch_version: ["2.5.1", "2.6.0", "2.7.1"]
     timeout-minutes: 20
 
     steps:
@@ -175,9 +175,9 @@ jobs:
 
       - name: Run tests
         run: |
-          pytest -v --durations=10 -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
-          pytest -v --durations=10 tests/patched/
-          pytest -v --durations=10 tests/cli/
+          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
+          pytest -v tests/patched/
+          pytest -v tests/cli/
 
       - name: cleanup pip cache
         run: |
@@ -198,7 +198,7 @@ jobs:
           - cuda: 126
             cuda_version: 12.6.3
             python_version: "3.11"
-            pytorch: 2.7.1
+            pytorch: 2.6.0
             num_gpus: 1
             axolotl_extras:
           - cuda: 126
@@ -252,6 +252,18 @@ jobs:
             python_version: "3.11"
             pytorch: 2.6.0
             num_gpus: 1
+            axolotl_extras: llmcompressor
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.5.1
+            num_gpus: 1
+            axolotl_extras:
+          - cuda: 126
+            cuda_version: 12.6.3
+            python_version: "3.11"
+            pytorch: 2.7.1
+            num_gpus: 1
             axolotl_extras:
           - cuda: 128
             cuda_version: 12.8.1

.pre-commit-config.yaml

@@ -36,7 +36,7 @@ repos:
             'pydantic>=2.5.3',
         ]
 -   repo: https://github.com/PyCQA/bandit
-    rev: 1.8.6
+    rev: 1.8.5
     hooks:
     -   id: bandit
         args: [

.runpod/src/config/config.yaml

@@ -97,7 +97,7 @@
 #       # 'no_input_format' cannot include {input}
 #       no_input_format: "{instruction} "
 
-#       # For `completion` datasets only, uses the provided field instead of `text` column
+#       # For `completion` datsets only, uses the provided field instead of `text` column
 #       field:
 
 # # Axolotl attempts to save the dataset as an arrow after packing the data together so

MANIFEST.in

@@ -2,5 +2,4 @@ include requirements.txt
 include README.md
 include LICENSE
 include src/setuptools_axolotl_dynamic_dependencies.py
-include src/axolotl/utils/chat_templates/templates/*.jinja
 recursive-include axolotl *.py

README.md

@@ -55,7 +55,7 @@ Features:
 
 - NVIDIA GPU (Ampere or newer for `bf16` and Flash Attention) or AMD GPU
 - Python 3.11
-- PyTorch ≥2.6.0
+- PyTorch ≥2.5.1
 
 ### Installation
 

_quarto.yml

@@ -276,7 +276,6 @@ website:
             - docs/torchao.qmd
             - docs/custom_integrations.qmd
             - docs/sequence_parallelism.qmd
-            - docs/gradient_checkpointing.qmd
 
         - section: "Troubleshooting"
           contents:

cicd/Dockerfile.jinja

@@ -9,7 +9,6 @@ ENV GITHUB_REF="{{ GITHUB_REF }}"
 ENV GITHUB_SHA="{{ GITHUB_SHA }}"
 ENV NIGHTLY_BUILD="{{ NIGHTLY_BUILD }}"
 ENV HF_HOME="{{ HF_HOME }}"
-ENV AXOLOTL_DATASET_PROCESSES="8"
 
 RUN apt-get update && \
     apt-get install -y --allow-change-held-packages vim curl nano libnccl2 libnccl-dev

cicd/multigpu.py

@@ -24,9 +24,9 @@ df_template = template_env.get_template("Dockerfile.jinja")
 df_args = {
     "AXOLOTL_EXTRAS": os.environ.get("AXOLOTL_EXTRAS", ""),
     "AXOLOTL_ARGS": os.environ.get("AXOLOTL_ARGS", ""),
-    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.6.0"),
-    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu126-2.6.0"),
-    "CUDA": os.environ.get("CUDA", "126"),
+    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.5.1"),
+    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu124-2.5.1"),
+    "CUDA": os.environ.get("CUDA", "124"),
     "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
     "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
     "CODECOV_TOKEN": os.environ.get("CODECOV_TOKEN", ""),

cicd/single_gpu.py

@@ -24,9 +24,9 @@ df_template = template_env.get_template(dockerfile)
 df_args = {
     "AXOLOTL_EXTRAS": os.environ.get("AXOLOTL_EXTRAS", ""),
     "AXOLOTL_ARGS": os.environ.get("AXOLOTL_ARGS", ""),
-    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.6.0"),
-    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu126-2.6.0"),
-    "CUDA": os.environ.get("CUDA", "126"),
+    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.5.1"),
+    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu124-2.5.1"),
+    "CUDA": os.environ.get("CUDA", "124"),
     "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
     "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
     "NIGHTLY_BUILD": os.environ.get("NIGHTLY_BUILD", ""),

docker/Dockerfile-base

@@ -22,9 +22,11 @@ RUN apt-get update \
     && mkdir /root/.conda \
     && bash Miniconda3-latest-Linux-x86_64.sh -b \
     && rm -f Miniconda3-latest-Linux-x86_64.sh \
-    && conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
+    && conda create -n "axolotl-py${PYTHON_VERSION}" python="${PYTHON_VERSION}" \
+    && conda init bash \
+    && echo "conda activate axolotl-py${PYTHON_VERSION}" >> ~/.bashrc
 
-ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
+ENV PATH="/root/miniconda3/envs/axolotl-py${PYTHON_VERSION}/bin:${PATH}"
 
 WORKDIR /workspace
 

docker/Dockerfile-base-next

@@ -22,9 +22,11 @@ RUN apt-get update \
     && mkdir /root/.conda \
     && bash Miniconda3-latest-Linux-x86_64.sh -b \
     && rm -f Miniconda3-latest-Linux-x86_64.sh \
-    && conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
+    && conda create -n "axolotl-py${PYTHON_VERSION}" python="${PYTHON_VERSION}" \
+    && conda init bash \
+    && echo "conda activate axolotl-py${PYTHON_VERSION}" >> ~/.bashrc
 
-ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
+ENV PATH="/root/miniconda3/envs/axolotl-py${PYTHON_VERSION}/bin:${PATH}"
 
 WORKDIR /workspace
 

docker/Dockerfile-base-nightly

@@ -22,9 +22,11 @@ RUN apt-get update \
     && mkdir /root/.conda \
     && bash Miniconda3-latest-Linux-x86_64.sh -b \
     && rm -f Miniconda3-latest-Linux-x86_64.sh \
-    && conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
+    && conda create -n "axolotl-py${PYTHON_VERSION}" python="${PYTHON_VERSION}" \
+    && conda init bash \
+    && echo "conda activate axolotl-py${PYTHON_VERSION}" >> ~/.bashrc
 
-ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
+ENV PATH="/root/miniconda3/envs/axolotl-py${PYTHON_VERSION}/bin:${PATH}"
 
 WORKDIR /workspace
 

docs/custom_integrations.qmd

@@ -7,7 +7,6 @@ toc-depth: 3
 ```{python}
 #| echo: false
 
-import os
 import re
 
 def process_readme(integration_name):
@@ -54,24 +53,6 @@ sections = [
     ("LLMCompressor", "llm_compressor")
 ]
 
-for folder_name in os.listdir("../src/axolotl/integrations/"):
-    if folder_name in [path for name, path in sections]:
-        # skip if already in sections
-        continue
-    if os.path.exists(f"../src/axolotl/integrations/{folder_name}/README.md"):
-        # grab the first heading in README.md as the section name
-        with open(f"../src/axolotl/integrations/{folder_name}/README.md", "r") as f:
-            txt = f.read()
-            matches = re.search(r'^# (.*)\n?', txt, flags=re.MULTILINE)
-            if matches:
-                name = matches.group(1)
-            else:
-                continue
-            sections.append((name, folder_name))
-
-# sort sections by name
-sections = sorted(sections, key=lambda x: x[0])
-
 for section_name, folder_name in sections:
     print(print_section(section_name, folder_name))
 ```

docs/dataset-formats/conversation.qmd

@@ -187,7 +187,6 @@ Instead of passing `tools` via the system prompt, an alternative method would be
             "role": "assistant", // call the function via assistant
             "tool_calls": [
                 {
-                    "id": "...",  // required only for mistral
                     "type": "function",
                     "function": {
                         "name": "...",
@@ -200,7 +199,6 @@ Instead of passing `tools` via the system prompt, an alternative method would be
         },
         {
             "role": "tool",
-            "tool_call_id": "...",  // required only for mistral
             "name": "...",
             "content": "..."
         },

docs/docker.qmd

@@ -34,9 +34,9 @@ Tags examples:
 
 - `main-base-py3.11-cu128-2.7.1`
 - `main-base-py3.11-cu126-2.7.1`
-- `main-base-py3.11-cu126-2.7.0`
 - `main-base-py3.11-cu126-2.6.0`
 - `main-base-py3.11-cu124-2.6.0`
+- `main-base-py3.11-cu124-2.5.1`
 
 ## Main
 
@@ -76,12 +76,12 @@ Tags examples:
 
 - `main-py3.11-cu128-2.7.1`
 - `main-py3.11-cu126-2.7.1`
-- `main-py3.11-cu126-2.7.0`
 - `main-py3.11-cu126-2.6.0`
 - `main-py3.11-cu124-2.6.0`
+- `main-py3.11-cu124-2.5.1`
 - `main-latest`
 - `main-20250303-py3.11-cu124-2.6.0`
-- `main-20250303-py3.11-cu126-2.6.0`
+- `main-20250303-py3.11-cu124-2.5.1`
 - `0.10.1`
 
 ## Cloud

docs/faq.qmd

@@ -55,14 +55,6 @@ description: Frequently asked questions
 
 > A: This is because you may be using `preprocess` CLI with `pretraining_dataset:` or `skip_prepare_dataset: true` respectively. Please use `axolotl train` CLI directly instead as these datasets are prepared on demand.
 
-**Q: vLLM is not working with Axolotl**
-
-> A: We currently recommend torch 2.6.0 for use with `vllm`. Please ensure you use the right version. For Docker, please use the `main-py3.11-cu124-2.6.0` tag.
-
-**Q: FA2 2.8.0 `undefined symbol` runtime error on CUDA 12.4**
-
-> A: There seems to be a wheel issue with FA2 2.8.0 on CUDA 12.4. Try CUDA 12.6 instead or downgrade to FA2 2.7.4. Please refer to the upstream issue: https://github.com/Dao-AILab/flash-attention/issues/1717.
-
 ### Chat templates
 
 **Q: `jinja2.exceptions.UndefinedError: 'dict object' has no attribute 'content' / 'role' / ____`**

docs/fsdp_qlora.qmd

@@ -20,7 +20,7 @@ To enable `QLoRA` with `FSDP`, you need to perform the following steps:
 > See the [example config](#example-config) file in addition to reading these instructions.
 
 1. Set `adapter: qlora` in your axolotl config file.
-2. Enable FSDP in your axolotl config, as [described here](multi-gpu.qmd#sec-fsdp).
+2. Enable FSDP in your axolotl config, as [described here](https://github.com/axolotl-ai-cloud/axolotl?tab=readme-ov-file#fsdp).
 3. Use one of the supported model types: `llama`, `mistral` or `mixtral`.
 
 ## Example Config

docs/gradient_checkpointing.qmd

@@ -1,29 +0,0 @@
----
-title: Gradient Checkpointing and Activation Offloading
----
-
-Gradient checkpointing and activation offloading are techniques used to optimize the performance of deep learning
-models by reducing the memory footprint and improving computational efficiency.
-
-### Enabling Gradient Checkpointing
-
-```yaml
-gradient_checkpointing: true
-```
-
-### Enabling Activation Offloading
-
-```yaml
-gradient_checkpointing: true  # required for activation offloading
-activation_offloading: true
-```
-
-Activation offloading variants:
-
-The default `activation_offloading: true` offloads activations to CPU and uses CUDA streams
-to overlap the communications and computations when offloading.
-
-The `activation_offloading: legacy` naively offloads activations to CPU and without additional optimizations.
-
-For resource constrained environments with limited CPU memory, `activation_offloading: disk` offloads
-activations to disk instead of CPU RAM so that much larger context lengths can be trained with minimal memory.

docs/installation.qmd

@@ -15,7 +15,7 @@ This guide covers all the ways you can install and set up Axolotl for your envir
 
 - NVIDIA GPU (Ampere architecture or newer for `bf16` and Flash Attention) or AMD GPU
 - Python ≥3.11
-- PyTorch ≥2.6.0
+- PyTorch ≥2.5.1
 
 ## Installation Methods {#sec-installation-methods}
 

docs/multi-gpu.qmd

@@ -23,6 +23,8 @@ Axolotl supports several methods for multi-GPU training:
 
 ## DeepSpeed {#sec-deepspeed}
 
+DeepSpeed is the recommended approach for multi-GPU training due to its stability and performance. It provides various optimization levels through ZeRO stages.
+
 ### Configuration {#sec-deepspeed-config}
 
 Add to your YAML config:
@@ -30,6 +32,7 @@ Add to your YAML config:
 ```{.yaml}
 deepspeed: deepspeed_configs/zero1.json
 ```
+
 ### Usage {#sec-deepspeed-usage}
 
 ```{.bash}
@@ -63,75 +66,9 @@ Start from Stage 1 -> Stage 2 -> Stage 3.
 
 :::
 
-::: {.callout-tip}
+## FSDP {#sec-fsdp}
 
-Using ZeRO Stage 3 with Single-GPU training
-
-ZeRO Stage 3 can be used for training on a single GPU by manually setting the environment variables:
-`WORLD_SIZE=1 LOCAL_RANK=0 MASTER_ADDR=0.0.0.0 MASTER_PORT=29500`
-
-:::
-
-## Fully Sharded Data Parallel (FSDP) {#sec-fsdp}
-
-::: {.callout-note}
-
-FSDP2 is recommended for new users. FSDP1 is deprecated and will be removed in an upcoming release of Axolotl.
-
-:::
-
-### Migrating from FSDP1 to FSDP2 {#sec-migrate-fsdp1-fsdp2}
-
-To migrate your config from FSDP1 to FSDP2, you must use the `fsdp_version` top-level config field to specify the FSDP version, and
-also follow the config field mapping below to update field names.
-
-#### Config mapping
-
-FSDP1 | FSDP2
--------- | --------
-fsdp_sharding_strategy | reshard_after_forward
-fsdp_backward_prefetch_policy | **REMOVED**
-fsdp_backward_prefetch | **REMOVED**
-fsdp_forward_prefetch | **REMOVED**
-fsdp_sync_module_states | **REMOVED**
-fsdp_cpu_ram_efficient_loading | cpu_ram_efficient_loading
-fsdp_state_dict_type | state_dict_type
-fsdp_use_orig_params | **REMOVED**
-
-
-For example, if you were using the following FSDP1 config:
-
-```{.yaml}
-fsdp_version: 1
-fsdp_config:
-  fsdp_offload_params: false
-  fsdp_cpu_ram_efficient_loading: true
-  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
-  fsdp_transformer_layer_cls_to_wrap: Qwen3DecoderLayer
-  fsdp_state_dict_type: FULL_STATE_DICT
-  fsdp_sharding_strategy: FULL_SHARD
-```
-
-You can migrate to the following FSDP2 config:
-
-```{.yaml}
-fsdp_version: 2
-fsdp_config:
-  offload_params: false
-  cpu_ram_efficient_loading: true
-  auto_wrap_policy: TRANSFORMER_BASED_WRAP
-  transformer_layer_cls_to_wrap: Qwen3DecoderLayer
-  state_dict_type: FULL_STATE_DICT
-  reshard_after_forward: true
-```
-
-### FSDP1 (deprecated) {#sec-fsdp-config}
-
-::: {.callout-note}
-
-Using `fsdp` to configure FSDP is deprecated and will be removed in an upcoming release of Axolotl. Please use `fsdp_config` as above instead.
-
-:::
+### Basic FSDP Configuration {#sec-fsdp-config}
 
 ```{.yaml}
 fsdp:
@@ -143,7 +80,6 @@ fsdp_config:
   fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
 ```
 
-
 ## Sequence parallelism {#sec-sequence-parallelism}
 
 We support sequence parallelism (SP) via the

docs/multi-node.qmd

@@ -40,13 +40,13 @@ use_cpu: false
 
 Configure your model to use FSDP in the Axolotl yaml. For example:
 ```yaml
-fsdp_version: 2
+fsdp:
+  - full_shard
+  - auto_wrap
 fsdp_config:
-  offload_params: true
-  state_dict_type: FULL_STATE_DICT
-  auto_wrap_policy: TRANSFORMER_BASED_WRAP
-  transformer_layer_cls_to_wrap: LlamaDecoderLayer
-  reshard_after_forward: true
+  fsdp_offload_params: true
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
 ```
 
 All you have to do now is launch using accelerate as you would usually do on each machine and voila, the processes will start once you have launched accelerate on every machine.

docs/rlhf.qmd

@@ -17,6 +17,7 @@ feedback. Various methods include, but not limited to:
 - [Kahneman-Tversky Optimization (KTO)](#kto)
 - [Odds Ratio Preference Optimization (ORPO)](#orpo)
 - [Group Relative Policy Optimization (GRPO)](#grpo)
+- Proximal Policy Optimization (PPO) (not yet supported in axolotl, if you're interested in contributing, please reach out!)
 
 
 ## RLHF using Axolotl
@@ -274,14 +275,15 @@ rl: dpo
 datasets:
   - path: ...
     split: train
-    type:
-      field_prompt: "prompt"
-      field_system: "system"
-      field_chosen: "chosen"
-      field_rejected: "rejected"
-      prompt_format: "{prompt}"
-      chosen_format: "{chosen}"
-      rejected_format: "{rejected}"
+    type: user_defined.default
+
+    field_prompt: "prompt"
+    field_system: "system"
+    field_chosen: "chosen"
+    field_rejected: "rejected"
+    prompt_format: "{prompt}"
+    chosen_format: "{chosen}"
+    rejected_format: "{rejected}"
 ```
 
 The input format is a simple JSON input with customizable fields based on the above config.
@@ -474,13 +476,14 @@ rl: kto
 datasets:
   - path: ...
     split: train
-    type:
-      field_prompt: "prompt"
-      field_system: "system"
-      field_completion: "completion"
-      field_label: "label"
-      prompt_format: "{prompt}"
-      completion_format: "{completion}"
+    type: user_defined.default
+
+    field_prompt: "prompt"
+    field_system: "system"
+    field_completion: "completion"
+    field_label: "label"
+    prompt_format: "{prompt}"
+    completion_format: "{completion}"
 ```
 
 The input format is a simple JSON input with customizable fields based on the above config.

examples/archived/README.md

@@ -1,5 +0,0 @@
-# Archived Examples
-
-This directory contains examples that are no longer maintained and may no longer be functional.
-
-We keep them around for archival purposes in case they are useful to others.

examples/cloud/modal.yaml

@@ -26,5 +26,3 @@ timeout: 86400
 # Preprocess specific configurations
 memory_preprocess: 32
 timeout_preprocess: 14400
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/cohere/command-r-7b-qlora.yml

@@ -35,6 +35,7 @@ wandb_watch:
 wandb_name:
 wandb_log_model:
 
+
 gradient_accumulation_steps: 4
 micro_batch_size: 1
 num_epochs: 4
@@ -55,5 +56,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/colab-notebooks/colab-axolotl-example.ipynb

@@ -40,7 +40,7 @@
         "%%capture\n",
         "# This step can take ~5-10 minutes to install dependencies\n",
         "!pip install --no-build-isolation axolotl[flash-attn]>=0.9.1\n",
-        "!pip install \"cut-cross-entropy[transformers] @ git+https://github.com/axolotl-ai-cloud/ml-cross-entropy.git@50cef19\""
+        "!pip install \"cut-cross-entropy[transformers] @ git+https://github.com/axolotl-ai-cloud/ml-cross-entropy.git@78b2a45713a54c9bedf8b33f5e31cf07a1a57154\""
       ]
     },
     {

examples/deepcogito/cogito-v1-preview-llama-3B-lora.yml

@@ -56,5 +56,3 @@ evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/deepcogito/cogito-v1-preview-qwen-14B-lora.yml

@@ -56,5 +56,3 @@ evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/deepseek-v2/fft-fsdp-16b.yaml

@@ -55,5 +55,3 @@ fsdp_config:
   fsdp_transformer_layer_cls_to_wrap: DeepseekV2DecoderLayer
   fsdp_state_dict_type: FULL_STATE_DICT
   fsdp_sharding_strategy: FULL_SHARD
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/deepseek-v2/qlora-fsdp-2_5.yaml

@@ -79,5 +79,3 @@ fsdp_config:
   fsdp_transformer_layer_cls_to_wrap: DeepseekV2DecoderLayer
   fsdp_state_dict_type: FULL_STATE_DICT
   fsdp_sharding_strategy: FULL_SHARD
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/devstral/README.md

@@ -1,70 +0,0 @@
-# Finetune Devstral with Axolotl
-
-Devstral Small is a 24B parameter opensource model from MistralAI found on HuggingFace [Devstral-Small-2505](https://huggingface.co/mistralai/Devstral-Small-2505) and [Devstral-Small-2507](https://huggingface.co/mistralai/Devstral-Small-2507). `Devstral-Small-2507` is the latest version of the model and has [function calling](https://mistralai.github.io/mistral-common/usage/tools/) support.
-
-This guide shows how to fine-tune it with Axolotl with multi-turn conversations with proper masking.
-
-The model was fine-tuned ontop of [Mistral-Small-3.1](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Base-2503) without the vision layer and has a context of up to 128k tokens.
-
-Thanks to the team at MistralAI for giving us early access to prepare for this release.
-
-## Getting started
-
-1. Install Axolotl following the [installation guide](https://docs.axolotl.ai/docs/installation.html). You need to install from main as Devstral is only on nightly or use our latest [Docker images](https://docs.axolotl.ai/docs/docker.html).
-
-    Here is an example of how to install from main for pip:
-
-```bash
-# Ensure you have Pytorch installed (Pytorch 2.6.0+)
-git clone https://github.com/axolotl-ai-cloud/axolotl.git
-cd axolotl
-
-pip3 install packaging==23.2 setuptools==75.8.0 wheel ninja
-pip3 install --no-build-isolation -e '.[flash-attn]'
-```
-
-2. Run the finetuning example:
-
-```bash
-axolotl train examples/devstral/devstral-small-qlora.yml
-```
-
-This config uses about 21GB VRAM.
-
-Let us know how it goes. Happy finetuning! 🚀
-
-### TIPS
-
-- You can run a full finetuning by removing the `adapter: qlora` and `load_in_4bit: true` from the config.
-- Read more on how to load your own dataset at [docs](https://docs.axolotl.ai/docs/dataset_loading.html).
-- The dataset format follows the OpenAI Messages format as seen [here](https://docs.axolotl.ai/docs/dataset-formats/conversation.html#chat_template).
-- Learn how to use function calling with Axolotl at [docs](https://docs.axolotl.ai/docs/dataset-formats/conversation.html#using-tool-use).
-
-## Optimization Guides
-
-- [Multi-GPU Training](https://docs.axolotl.ai/docs/multi-gpu.html)
-- [Multi-Node Training](https://docs.axolotl.ai/docs/multi-node.html)
-- [LoRA Optimizations](https://docs.axolotl.ai/docs/lora_optims.html)
-- [Cut Cross Entropy](https://docs.axolotl.ai/docs/custom_integrations.html#cut-cross-entropy)
-- [Liger Kernel](https://docs.axolotl.ai/docs/custom_integrations.html#liger-kernels)
-
-## Limitations
-
-We only support the `mistral-common` tokenizer for Supervised Fine-tuning at the moment and for `type: chat_template` only.
-
-In addition, we do not support overriding tokens yet.
-
-## Related Resources
-
-- [MistralAI Devstral Blog](https://mistral.ai/news/devstral)
-- [MistralAI Devstral 1.1 Blog](https://mistral.ai/news/devstral-2507)
-- [Axolotl Docs](https://docs.axolotl.ai)
-- [Axolotl GitHub](https://github.com/axolotl-ai-cloud/axolotl)
-- [Axolotl Website](https://axolotl.ai)
-- [Axolotl Discord](https://discord.gg/7m9sfhzaf3)
-
-
-## Future Work
-
-- Add parity to Preference Tuning, RL, Multi-modal, etc.
-- Add parity to other tokenizer configs like overriding tokens.

examples/devstral/devstral-small-qlora.yml

@@ -1,66 +0,0 @@
-base_model: mistralai/Devstral-Small-2507
-
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-# Enable to use mistral-common tokenizer
-tokenizer_use_mistral_common: true
-
-load_in_8bit: false
-load_in_4bit: true
-
-plugins:
-  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
-
-datasets:
-  - path: fozziethebeat/alpaca_messages_2k_test
-    type: chat_template
-
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.1
-output_dir: ./outputs/qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 2048
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0
-lora_target_linear: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_torch
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: false
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-loss_watchdog_threshold: 5.0
-loss_watchdog_patience: 3
-
-warmup_ratio: 0.05
-evals_per_epoch: 4
-saves_per_epoch: 1
-
-weight_decay: 0.0
-special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-1b-deep-qlora.yaml

@@ -69,5 +69,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-1b-qlora.yaml

@@ -46,6 +46,7 @@ wandb_watch:
 wandb_name:
 wandb_log_model:
 
+
 gradient_accumulation_steps: 4
 micro_batch_size: 1
 num_epochs: 4
@@ -68,5 +69,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-34b-qlora.yaml

@@ -69,5 +69,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-3b-qlora.yaml

@@ -69,5 +69,3 @@ evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-500m-qlora.yaml

@@ -69,5 +69,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/falcon-h1/falcon-h1-7b-qlora.yaml

@@ -69,5 +69,3 @@ evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/gemma2/qlora.yml

@@ -60,5 +60,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/gemma2/reward-model.yaml

@@ -50,5 +50,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/gemma3/gemma-3-1b-qlora.yml

@@ -66,5 +66,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/gemma3/gemma-3-4b-qlora.yml

@@ -60,5 +60,3 @@ warmup_ratio: 0.1
 evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/gemma3/gemma-3-4b-vision-qlora.yml

@@ -62,5 +62,3 @@ warmup_ratio: 0.1
 evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/glm4/qlora-32b.yaml

@@ -60,5 +60,3 @@ evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/jamba/qlora.yaml

@@ -54,5 +54,3 @@ evals_per_epoch:
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/jamba/qlora_deepspeed.yaml

@@ -55,5 +55,3 @@ saves_per_epoch: 1
 deepspeed: deepspeed_configs/zero2.json
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/jamba/qlora_fsdp_large.yaml

@@ -64,5 +64,3 @@ fsdp_config:
   fsdp_transformer_layer_cls_to_wrap: JambaAttentionDecoderLayer,JambaMambaDecoderLayer
   fsdp_state_dict_type: FULL_STATE_DICT
   fsdp_sharding_strategy: FULL_SHARD
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/lfm2/README.md

@@ -1,7 +0,0 @@
-# Liquid Foundation Models 2
-
-LFM2 support in transformers exists in the main branch, but is not yet included in the transformers release.
-
-```bash
-pip install --upgrade --no-deps --force-reinstall git+https://github.com/huggingface/transformers.git
-```

examples/lfm2/lfm2-350m-fft.yaml

@@ -1,50 +0,0 @@
-base_model: LiquidAI/LFM2-350M
-
-chunked_cross_entropy: true
-
-chat_template: tokenizer_default
-eot_tokens:
-  - "<|im_end|>"
-datasets:
-  - path: mlabonne/FineTome-100k
-    type: chat_template
-    split: train[:20%]
-    field_messages: conversations
-    message_field_role: from
-    message_field_content: value
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.05
-output_dir: ./outputs/out
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 2
-micro_batch_size: 4
-num_epochs: 1
-optimizer: adamw_torch_fused
-lr_scheduler: cosine
-learning_rate: 5e-5
-
-bf16: true
-tf32: true
-
-gradient_checkpointing: false
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_ratio: 0.1
-evals_per_epoch: 2
-saves_per_epoch: 1
-
-weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/fft_optimized.yml

@@ -55,5 +55,3 @@ saves_per_epoch: 1
 deepspeed: #deepspeed_configs/zero2.json # multi-gpu only
 weight_decay: 0.1
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/gptq-lora.yml

@@ -64,5 +64,3 @@ special_tokens:
   bos_token: "<s>"
   eos_token: "</s>"
   unk_token: "<unk>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/lisa.yml

@@ -60,5 +60,3 @@ special_tokens:
   bos_token: "<s>"
   eos_token: "</s>"
   unk_token: "<unk>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/loftq.yml

@@ -52,5 +52,3 @@ evals_per_epoch: 4
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/lora.yml

@@ -52,5 +52,3 @@ evals_per_epoch: 4
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/qlora-fsdp.yml

@@ -67,5 +67,3 @@ fsdp_config:
   fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
   fsdp_state_dict_type: FULL_STATE_DICT
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/qlora.yml

@@ -53,5 +53,3 @@ evals_per_epoch: 4
 saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-2/relora.yml

@@ -58,5 +58,3 @@ special_tokens:
   bos_token: "<s>"
   eos_token: "</s>"
   unk_token: "<unk>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3-vision/lora-11b.yaml

@@ -57,5 +57,3 @@ warmup_ratio: 0.1
 evals_per_epoch: 1
 saves_per_epoch: 1
 weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/3b-qat-fsdp2.yaml

@@ -77,5 +77,3 @@ fsdp_config:
 
 special_tokens:
   pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/fft-8b-liger-fsdp.yaml

@@ -72,5 +72,3 @@ fsdp_config:
 special_tokens:
   pad_token: <|finetune_right_pad_id|>
   eos_token: <|eot_id|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/fft-8b.yaml

@@ -42,5 +42,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/instruct-dpo-lora-8b.yml

@@ -71,5 +71,3 @@ warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
 weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/instruct-lora-8b.yml

@@ -64,5 +64,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
    pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b-deduplicate-dpo.yml

@@ -83,5 +83,3 @@ warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
 weight_decay: 0.0
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b-deduplicate-sft.yml

@@ -61,5 +61,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
    pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b-kernels.yml

@@ -65,5 +65,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: "<|end_of_text|>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b-ray.yml

@@ -64,5 +64,3 @@ special_tokens:
 
 use_ray: true
 ray_num_workers: 4
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b-sample-packing-sequentially.yml

@@ -63,5 +63,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-1b.yml

@@ -60,5 +60,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: "<|end_of_text|>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/lora-8b.yml

@@ -57,5 +57,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
    pad_token: <|end_of_text|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/qlora-1b-kto.yaml

@@ -61,5 +61,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: "<|end_of_text|>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/qlora-1b.yml

@@ -62,5 +62,3 @@ saves_per_epoch: 1
 weight_decay: 0.0
 special_tokens:
   pad_token: "<|end_of_text|>"
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

examples/llama-3/qlora-fsdp-405b.yaml

@@ -60,5 +60,3 @@ fsdp_config:
   fsdp_sharding_strategy: FULL_SHARD
 special_tokens:
   pad_token: <|finetune_right_pad_id|>
-
-# save_first_step: true  # uncomment this to validate checkpoint saving works with your config