Spaces:
Sleeping
Sleeping
Prasham.Jain Claude Sonnet 4.6 commited on
Commit ·
15e36fe
1
Parent(s): 8580936
fix(training): upgrade base image to torch 2.6.0+cu126
Browse filesunsloth-zoo 2026.4.x requires torchao>=0.13.0 which uses torch.int1 —
a dtype added in PyTorch 2.6.0. The previous 2.5.1 image caused a hard
import failure cascading through transformers→peft→trl.
Changes:
- FROM pytorch/pytorch:2.5.1-cuda12.4 → pytorch/pytorch:2.6.0-cuda12.6
- unsloth extras: cu124-torch251 → cu126-torch260
- Remove torchao==0.5.0 pin (no longer needed, torch 2.6 supports 0.13+)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Dockerfile.train +10 -14
Dockerfile.train
CHANGED
|
@@ -9,8 +9,9 @@
|
|
| 9 |
# HF_SCENARIOS_REPO, HF_SFT_DATASET_REPO, HF_MODEL_REPO (optional)
|
| 10 |
# GRPO_STEPS (optional, default 100)
|
| 11 |
|
| 12 |
-
# torch 2.
|
| 13 |
-
|
|
|
|
| 14 |
|
| 15 |
ENV DEBIAN_FRONTEND=noninteractive
|
| 16 |
ENV PYTHONUNBUFFERED=1
|
|
@@ -21,26 +22,21 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
|
|
| 21 |
|
| 22 |
WORKDIR /workspace
|
| 23 |
|
| 24 |
-
#
|
| 25 |
-
#
|
| 26 |
-
#
|
| 27 |
RUN pip install --no-cache-dir \
|
| 28 |
-
"unsloth[
|
| 29 |
|
| 30 |
-
#
|
| 31 |
-
# unsloth pulls torchao>=0.8 which references torch.int1 (added in torch 2.6.0).
|
| 32 |
-
# Forcing 0.5.0 keeps the import chain clean; we use bitsandbytes, not torchao.
|
| 33 |
-
RUN pip install --no-cache-dir "torchao==0.5.0" --force-reinstall
|
| 34 |
-
|
| 35 |
-
# 2. Install project deps (unsloth already locked transformers/trl/peft above).
|
| 36 |
COPY pyproject.toml README.md ./
|
| 37 |
COPY src/ src/
|
| 38 |
RUN pip install --no-cache-dir -e ".[data,training]"
|
| 39 |
|
| 40 |
-
#
|
| 41 |
RUN pip install --no-cache-dir jupyterlab ipywidgets
|
| 42 |
|
| 43 |
-
#
|
| 44 |
COPY notebooks/ notebooks/
|
| 45 |
COPY train.py ./
|
| 46 |
|
|
|
|
| 9 |
# HF_SCENARIOS_REPO, HF_SFT_DATASET_REPO, HF_MODEL_REPO (optional)
|
| 10 |
# GRPO_STEPS (optional, default 100)
|
| 11 |
|
| 12 |
+
# torch 2.6.0 + CUDA 12.6 — minimum required by unsloth-zoo 2026.4.x
|
| 13 |
+
# (torchao>=0.13 needed by unsloth-zoo uses torch.int1, added in torch 2.6.0)
|
| 14 |
+
FROM pytorch/pytorch:2.6.0-cuda12.6-cudnn9-devel
|
| 15 |
|
| 16 |
ENV DEBIAN_FRONTEND=noninteractive
|
| 17 |
ENV PYTHONUNBUFFERED=1
|
|
|
|
| 22 |
|
| 23 |
WORKDIR /workspace
|
| 24 |
|
| 25 |
+
# Install unsloth for torch 2.6.0 + CUDA 12.6.
|
| 26 |
+
# This pulls compatible: transformers>=4.51 (Qwen3), peft, trl, accelerate,
|
| 27 |
+
# xformers, and torchao>=0.13 (all compatible with torch 2.6.0).
|
| 28 |
RUN pip install --no-cache-dir \
|
| 29 |
+
"unsloth[cu126-torch260] @ git+https://github.com/unslothai/unsloth.git"
|
| 30 |
|
| 31 |
+
# Install project deps (transformers/trl/peft already resolved by unsloth above).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
COPY pyproject.toml README.md ./
|
| 33 |
COPY src/ src/
|
| 34 |
RUN pip install --no-cache-dir -e ".[data,training]"
|
| 35 |
|
| 36 |
+
# JupyterLab for interactive mode
|
| 37 |
RUN pip install --no-cache-dir jupyterlab ipywidgets
|
| 38 |
|
| 39 |
+
# Copy notebooks and training scripts
|
| 40 |
COPY notebooks/ notebooks/
|
| 41 |
COPY train.py ./
|
| 42 |
|