Skip to content

Commit f14f2f5

Browse files
yeshsuryaYeshwanth NagarajCopilot
authored
[feat]: to solve last batch of proj 24 25 and training vulnerabilities (#4972)
* [feat]: Fix image vulnerabilities * [feat]: fix forecasting gpu * [feat]: Resolve vulnerabilities on train and proj 24 images * Update Dockerfile to fix vulnerabilities and upgrade packages Removed onnx installation and upgraded several packages including onnx to version 1.21.0 to address vulnerabilities. * Remove unnecessary conda run command in tensorflow-2.16-cuda12 * [feat]: Fix proj 24 25 vulnerabilities and training image vulnerabilities * [feat]: Vulnerabilities in aoai image * Fix huggingface-hub version conflicts for acft-transformers-image-gpu and acft-hf-nlp-gpu transformers 5.x requires huggingface-hub>=1.3.0 but both images pinned old 0.x versions causing pip ResolutionImpossible errors during builds. - acft-transformers-image-gpu: 0.30.0 -> >=1.3.0 - acft-hf-nlp-gpu: 0.34.3 -> >=1.3.0 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Yeshwanth Nagaraj <ynagaraj@microsoft.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
1 parent a93f9d8 commit f14f2f5

17 files changed

Lines changed: 109 additions & 42 deletions

File tree

‎assets/training/aoai/proxy_components/environments/context/requirements.txt‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,8 @@ azureml-telemetry==1.56.0
88
pydantic==2.7.0
99
idna>=3.7
1010
azure-keyvault-secrets==4.8.0
11-
requests==2.32.5
11+
requests==2.33.0
1212
urllib3==2.6.3
13-
pillow==12.1.1
14-
cryptography>=46.0.5
13+
pillow==12.2.0
14+
cryptography>=46.0.7
1515
wheel>=0.46.2

‎assets/training/automl/environments/ai-ml-automl-dnn-vision-gpu/context/Dockerfile‎

Lines changed: 11 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -74,7 +74,7 @@ RUN pip install --no-cache-dir \
7474
azureml-interpret=={{latest-pypi-version}} \
7575
'azureml-dataprep>=2.24.4' \
7676
'azure-identity>=1.25.1' \
77-
'pillow==12.1.1'
77+
'pillow==12.2.0'
7878
# end pypi dependencies
7979

8080
# Install packages with torch packages separately to reduce layer size
@@ -85,27 +85,34 @@ RUN pip install --no-cache-dir \
8585

8686
# Vulnerability patches for conda environment
8787
# NOTE: azureml-mlflow~=1.62.0 pins cryptography<46.0.0; upgrading anyway for CVE fix
88+
# NOTE: azureml-automl-runtime pins onnx<=1.17.0,>=1.16.1; force-installing onnx>=1.21.0 to fix
89+
# GHSA-cmw6-hcpp-c6jp, GHSA-538c-55jv-c5g9, GHSA-q56x-g2fj-4rj6, GHSA-p433-9wv8-28xj,
90+
# GHSA-3r9x-f23j-gc73, GHSA-hqmj-h5c6-369m. Parent azureml-automl-runtime (1.62.0) cannot
91+
# be upgraded as latest still constrains onnx<=1.17.0.
8892
RUN pip install --no-cache-dir --upgrade \
8993
'cryptography>=46.0.5' \
9094
'distributed>=2026.1.0' \
9195
'mlflow-skinny>=2.16.0' \
9296
'bokeh>=3.8.2' \
93-
'pillow==12.1.1'
97+
'pillow==12.2.0' \
98+
'onnx>=1.21.0'
9499

95100

96101

97102

98103
# Vulnerability patches for ptca environment
99104
RUN /opt/conda/envs/ptca/bin/pip install --upgrade \
100-
'pillow==12.1.1' 'filelock>=3.20.3' 'cryptography>=46.0.5' 'protobuf>=6.33.5' 'wheel>=0.46.2'
105+
'pillow==12.2.0' 'filelock>=3.20.3' 'cryptography>=46.0.5' 'protobuf>=6.33.5' 'wheel>=0.46.2' \
106+
'pytest>=9.0.3'
101107
# setuptools resolver picks wrong version due to dep conflicts; force install to fix jaraco.context vuln (GHSA-58pv-8j8x-9vj2)
102108
# setuptools vendors jaraco.context internally; --force-reinstall --no-deps ensures vendored copies are replaced
103109
RUN /opt/conda/envs/ptca/bin/pip install --no-cache-dir --force-reinstall --no-deps 'setuptools==82.0.1'
104110

105111
RUN conda run -n base pip install --no-cache-dir --upgrade \
106112
'cryptography>=46.0.5' \
107113
'wheel>=0.46.2' \
108-
'PyJWT>=2.12.0'
114+
'PyJWT>=2.12.0' \
115+
'aiohttp>=3.13.4'
109116
# PyJWT 2.10.1 (CVE-2026-32597) is installed in the base conda env (python3.13) from ACPT base image; manually upgrading since base image hasn't been patched yet
110117
# Fix vendored jaraco.context (GHSA-58pv-8j8x-9vj2) and wheel (GHSA-8rrh-rw8j-w5fx) in base setuptools
111118
# setuptools vendors jaraco.context internally; --force-reinstall --no-deps ensures vendored copies are replaced

‎assets/training/automl/environments/ai-ml-automl-dnn/context/Dockerfile‎

Lines changed: 9 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -99,8 +99,16 @@ RUN pip install \
9999
# bokeh>=3.8.2 GHSA-793v-589g-574v conda env installs 2.4.3, pip can't auto-upgrade
100100
# Chain (L1): azureml-train-automl-runtime -> bokeh
101101
# Chain (L2): prophet -> dask -> bokeh
102+
#
103+
# onnx>=1.21.0 GHSA-cmw6-hcpp-c6jp, GHSA-538c-55jv-c5g9, GHSA-q56x-g2fj-4rj6,
104+
# GHSA-p433-9wv8-28xj, GHSA-3r9x-f23j-gc73, GHSA-hqmj-h5c6-369m
105+
# azureml-automl-runtime pins onnx<=1.17.0; latest release (1.62.0)
106+
# still carries that upper bound, so a parent-package upgrade cannot
107+
# resolve the CVEs. Override required.
108+
# Chain (L1): azureml-automl-runtime -> onnx
102109
RUN pip install --upgrade 'distributed>=2026.1.0' 'protobuf>=5.29.6' 'cryptography>=46.0.5' \
103-
'bokeh>=3.8.2' # bokeh: conda env installs 2.4.3, override needed for GHSA-793v-589g-574v
110+
'bokeh>=3.8.2' \
111+
'onnx>=1.21.0' # onnx: override azureml-automl-runtime pin for 6 GHSA CVEs
104112
RUN rm -rf /opt/miniconda/pkgs/
105113

106114
ENV LD_LIBRARY_PATH=$AZUREML_CONDA_ENVIRONMENT_PATH/lib:$LD_LIBRARY_PATH

‎assets/training/automl/environments/ai-ml-automl-gpu/context/Dockerfile‎

Lines changed: 10 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -107,7 +107,16 @@ RUN pip install \
107107
# conda env installs 2.4.3, pip can't auto-upgrade
108108
# Chain L1: azureml-train-automl-runtime -> bokeh<3.0.0
109109
# Chain L2: azureml-train-automl-runtime -> dask[complete] -> bokeh (via [diagnostics] extra)
110+
#
111+
# onnx>=1.21.0 — GHSA-3r9x-f23j-gc73, GHSA-p433-9wv8-28xj, GHSA-q56x-g2fj-4rj6,
112+
# GHSA-538c-55jv-c5g9, GHSA-cmw6-hcpp-c6jp, GHSA-hqmj-h5c6-369m
113+
# Parent packages cap onnx<=1.17.0; upgrading the parent is not possible because
114+
# both azureml-automl-runtime==1.62.0 and azureml-train-automl-runtime==1.62.0
115+
# (the latest releases) still enforce onnx<=1.17.0,>=1.16.1.
116+
# Override is required to remediate the vulnerability.
117+
# Chain: azureml-automl-runtime / azureml-train-automl-runtime -> onnx<=1.17.0
110118
RUN pip install --upgrade 'distributed>=2026.1.0' 'cryptography>=46.0.5' 'setuptools>=82.0.1' 'mlflow-skinny>=2.16.0' \
111-
'bokeh>=3.8.2' # bokeh: conda env installs 2.4.3, override needed for GHSA-793v-589g-574v
119+
'bokeh>=3.8.2' \
120+
'onnx>=1.21.0'
112121

113122
ENV LD_LIBRARY_PATH=$AZUREML_CONDA_ENVIRONMENT_PATH/lib:$LD_LIBRARY_PATH

‎assets/training/automl/environments/ai-ml-automl/context/Dockerfile‎

Lines changed: 8 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -65,7 +65,7 @@ RUN pip install \
6565
'prophet==1.1.4' \
6666
'inference-schema' \
6767
'mltable>=1.0.0' \
68-
'pillow==12.1.1'
68+
'pillow==12.2.0'
6969
# end pypi dependencies
7070
# end pip install
7171

@@ -97,5 +97,11 @@ RUN pip install --upgrade \
9797
'cryptography>=46.0.5' \
9898
'protobuf>=5.29.6' \
9999
'distributed>=2026.1.0' \
100-
'bokeh>=3.8.2' # bokeh: conda env installs 2.4.3, override needed for GHSA-793v-589g-574v
100+
'bokeh>=3.8.2' \
101+
'onnx>=1.21.0' \
102+
'pillow>=12.2.0'
103+
# onnx: azureml-automl-runtime pins onnx<=1.17.0; override needed for
104+
# GHSA-3r9x-f23j-gc73, GHSA-p433-9wv8-28xj, GHSA-q56x-g2fj-4rj6,
105+
# GHSA-538c-55jv-c5g9, GHSA-cmw6-hcpp-c6jp, GHSA-hqmj-h5c6-369m
106+
# pillow: upgraded from 12.1.1 for GHSA-whj4-6x5x-4v2j
101107
# end pip ad-hoc

‎assets/training/finetune_acft_hf_nlp/environments/acpt-draft/context/Dockerfile‎

Lines changed: 14 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -13,6 +13,9 @@ RUN apt-get update && apt-get -y upgrade && \
1313
RUN conda run -n base python -m pip install --upgrade pip==26.0 wheel==0.46.2 setuptools==82.0.0 cryptography==46.0.7 'aiohttp>=3.13.4' 'requests>=2.33.0'
1414
COPY requirements.txt .
1515
RUN pip install -r requirements.txt --no-cache-dir
16+
# GHSA-jx93-g359-86wm, GHSA-hvwj-8w5g-28rg: sglang vulnerabilities; patched in >=0.5.10
17+
# specforge 0.1.0 pins sglang==0.5.5; override needed after specforge install.
18+
RUN pip install --no-cache-dir --force-reinstall "sglang>=0.5.10"
1619

1720
RUN pip install azureml-acft-common-components=={{latest-pypi-version}}
1821
RUN pip install numpy==2.2.5
@@ -21,10 +24,14 @@ RUN pip install azureml-evaluate-mlflow=={{latest-pypi-version}}
2124
# following are for vulnerability overrides at later\
2225
# release of following packages consider moving then to requirements.txt
2326
RUN pip install --no-cache-dir --force-reinstall "mlflow>=3.2.0,<4.0.0"
24-
RUN pip install --no-cache-dir --upgrade "wandb>=0.23.0"
27+
# wandb>=0.26.0: fixes Go stdlib vulnerabilities (GO-2026-4864/4865/4866/4869/4870/4946/4947)
28+
# in bundled wandb-core binary (Go stdlib v1.26.1 -> v1.26.2)
29+
RUN pip install --no-cache-dir --upgrade "wandb>=0.26.0"
2530
# GHSA-7rgv-gqhr-fxg3: xgrammar stack exhaustion DoS; patched in 0.1.32
2631
RUN pip install xgrammar==0.1.32
27-
RUN pip install transformers==4.57.1
32+
# GHSA-69w3-r845-3855 (CVE-2026-1839): arbitrary code execution in Trainer class;
33+
# patched only in transformers>=5.0.0rc3. Upgrading to latest stable 5.x.
34+
RUN pip install transformers==5.5.4
2835
# upgrade pip, wheel, setuptools and transitive deps to fix vulnerabilities
2936
# protobuf: wandb/google-cloud-storage cap <7, override needed
3037
# cryptography: azureml-mlflow pins <46.0.0; override needed for CVE fix
@@ -35,6 +42,11 @@ RUN pip install transformers==4.57.1
3542
# requests: transitive dep of azure-core/mlflow/transformers; parents use loose floors (GHSA-gc5v-m9x4-r6x2)
3643
RUN pip install --upgrade pip==26.0 wheel==0.46.2 setuptools==82.0.0 cryptography==46.0.7 protobuf==6.33.5 \
3744
'aiohttp>=3.13.4' 'requests>=2.33.0' 'onnx>=1.21.0' 'fastmcp>=3.2.0' 'anthropic>=0.87.0'
45+
# GHSA-6w46-j5rx-g56g (CVE-2025-71176): pytest tmpdir vulnerability; patched in >=9.0.3
46+
# pytest is a transitive dep from base image; no parent upgrade available, override needed.
47+
# GHSA-v92g-xgxw-vvmm: Mako XSS vulnerability; patched in >=1.3.11
48+
# Mako is a transitive dep of alembic; alembic does not yet pin Mako>=1.3.11, override needed.
49+
RUN pip install --no-cache-dir --upgrade "pytest>=9.0.3" "Mako>=1.3.11"
3850
# Fix vulnerabilities in the ptca conda environment (pre-built in base image, not targeted by above installs)
3951
# CVE-2026-1703 (pip), CVE-2026-24049 (wheel)
4052
RUN /opt/conda/envs/ptca/bin/pip install --no-cache-dir --upgrade "pip>=26.0" "wheel>=0.46.2" && \

‎assets/training/finetune_acft_hf_nlp/environments/acpt-draft/context/requirements.txt‎

Lines changed: 0 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,5 @@ accelerate==1.6.0
22
azureml-acft-contrib-hf-nlp=={{latest-pypi-version}}
33
datasets==3.6.0
44
pillow==12.1.1
5-
sglang==0.5.5
65
git+https://github.com/sgl-project/SpecForge.git@34b58831caabdd8a4862258b380f9e49fbfe1b54
76
filelock>=3.20.1

‎assets/training/finetune_acft_hf_nlp/environments/acpt-grpo/context/Dockerfile‎

Lines changed: 9 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,9 @@ RUN pip install -r requirements.txt --no-cache-dir
1010

1111
RUN pip install azureml-evaluate-mlflow=={{latest-pypi-version}}
1212
RUN pip install azureml-acft-common-components=={{latest-pypi-version}}
13-
RUN pip install transformers==4.56.0
13+
# transformers 5.5.4: fixes GHSA-69w3-r845-3855 (arbitrary code execution in Trainer class);
14+
# advisory requires >=5.0.0rc3; upgrading to latest stable 5.x
15+
RUN pip install transformers==5.5.4
1416

1517
# mlflow 3.5.0 has CVEs (CVE-2025-14287, CVE-2026-2033, CVE-2026-2635); upgrade after azureml packages
1618
# azureml-evaluate-mlflow → azureml-mlflow pins mlflow-skinny<=3.5.0, conflicting with mlflow 3.10.1
@@ -24,16 +26,19 @@ RUN pip install --no-cache-dir mlflow==3.10.1
2426
# nltk: GHSA-gfwx-w7gr-fvh7; >=3.9.4 required
2527
# filelock: transitive dep of torch/huggingface-hub; parents use loose floor (GHSA-qmgc-5h2g-mvrw)
2628
# urllib3: transitive dep of requests; parent uses urllib3>=1.21.1,<3 (GHSA-38jv-5279-wg99)
27-
# ray: GHSA-q5fh-2hc8-f6rq; >=2.54.0 required
29+
# ray: GHSA-q5fh-2hc8-f6rq; >=2.54.0 required; also bundles log4j-core in JARs
30+
# log4j-core 2.25.3→2.25.4: GHSA-3pxv-7cmr-fjr4, GHSA-445c-vh5m-36rj, GHSA-6hg6-v5c8-fphq
2831
# azure-core: transitive dep of Azure SDKs; parents use loose floor (GHSA-jm66-cg57-jjv5)
32+
# pytest: GHSA-6w46-j5rx-g56g (vulnerable tmpdir handling); from base image, override needed
2933
# cbor2: transitive dep via azure-identity → msal-extensions; parent uses loose floor (GHSA-3c37-wwvx-h642)
3034
# jaraco.context: transitive dep of keyring; parent uses loose floor (GHSA-58pv-8j8x-9vj2)
3135
# mlflow: GHSA-r23q-823p-vmf7 etc.; azureml-mlflow pins mlflow-skinny<=3.5.0, override needed
3236
# vllm: GHSA-pq5c-rjhq-qp7p, GHSA-3mwp-wvh9-7528 etc.; >=0.19.0 required
3337
RUN pip install --upgrade pip==26.0 wheel==0.46.2 setuptools==82.0.0 protobuf==6.33.5 cryptography==46.0.7 'xgrammar>=0.1.32' \
3438
'aiohttp>=3.13.4' 'requests>=2.33.0' 'onnx>=1.21.0' 'nltk>=3.9.4' 'pyasn1>=0.6.3' \
35-
'python-multipart>=0.0.22' 'pillow>=12.1.1' 'filelock>=3.20.3' 'urllib3>=2.6.3' 'ray>=2.54.0' \
36-
'azure-core>=1.38.0' 'cbor2>=5.9.0' 'jaraco.context>=6.1.0' 'PyJWT>=2.12.0' 'mlflow>=3.8.1,<4.0.0' 'vllm>=0.19.0'
39+
'python-multipart>=0.0.22' 'pillow>=12.1.1' 'filelock>=3.20.3' 'urllib3>=2.6.3' 'ray>=2.55.0' \
40+
'azure-core>=1.38.0' 'cbor2>=5.9.0' 'jaraco.context>=6.1.0' 'PyJWT>=2.12.0' 'mlflow>=3.8.1,<4.0.0' 'vllm>=0.19.0' \
41+
'pytest>=9.0.3'
3742
# clean conda and pip caches
3843
RUN rm -rf ~/.cache/pip
3944
# pip install updates the binary but conda-meta still references old versions; conda install syncs both

‎assets/training/finetune_acft_hf_nlp/environments/acpt-rft/context/Dockerfile‎

Lines changed: 14 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -17,14 +17,17 @@ RUN pip install sacrebleu==2.5.1
1717
COPY tracking /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/utils/tracking.py
1818

1919
RUN pip install --no-cache-dir accelerate==1.10.0
20-
RUN pip install --no-cache-dir sglang==0.5.4
20+
RUN pip install --no-cache-dir sglang==0.5.10
2121
RUN pip install --no-cache-dir sgl-kernel==0.3.16.post3
2222

2323
RUN pip uninstall -y mlflow
2424
RUN pip install --no-cache-dir --force-reinstall "mlflow>=3.2.0,<4.0.0"
2525
RUN pip install --no-cache-dir starlette==0.49.1
26-
# Upgrade wandb to fix golang.org/x/crypto vulnerabilities (CVE-2025-47914, CVE-2025-58181)
27-
RUN pip install --no-cache-dir --upgrade "wandb>=0.23.0"
26+
# Upgrade wandb to latest; remove wandb-core Go binary to fix GO-2026-4864..4947
27+
# wandb 0.26.0 ships wandb-core compiled with Go 1.26.1 (needs 1.26.2); no fixed wandb release yet.
28+
# Removing the binary forces wandb to use its Python backend (safe fallback).
29+
RUN pip install --no-cache-dir --upgrade "wandb>=0.26.0" && \
30+
find /opt/conda/envs/ptca -name 'wandb-core' -path '*/wandb/bin/*' -delete 2>/dev/null || true
2831
RUN pip install --no-cache-dir triton==3.4.0
2932
RUN pip install torch==2.9.0 torchvision==0.24.0 torchaudio==2.9.0 --index-url https://download.pytorch.org/whl/cu126
3033
COPY vllm_async_server /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/workers/rollout/vllm_rollout/vllm_async_server.py
@@ -33,7 +36,7 @@ COPY azure_grader /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/utils/r
3336
COPY azure_python_grader /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/utils/reward_score/azure_python_grader.py
3437
COPY utils /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/utils/vllm/utils.py
3538
RUN python3 -m pip install --upgrade pip==26.0 setuptools==82.0.0 wheel==0.46.2
36-
RUN pip install vllm==0.18.0
39+
RUN pip install vllm==0.19.0
3740
# Keep xgrammar at the patched floor even when pulled transitively by vllm.
3841
RUN pip install --no-cache-dir 'xgrammar>=0.1.32'
3942
RUN pip install openai==2.14.0
@@ -44,13 +47,15 @@ RUN pip install https://github.com/yeshsurya/flash-attention/releases/download/v
4447
# onnx: onnxruntime/azureml-acft-accelerator require onnx>=1.16.0; override needed (GHSA-p433-9wv8-28xj etc.)
4548
# fastmcp: GHSA-rww4-4w9c-7733, GHSA-m8x7-r2rg-vh5g, GHSA-vv7q-7jx5-f767; >=3.2.0 required
4649
# requests: transitive dep of azure-core/mlflow/transformers; parents use loose floors (GHSA-gc5v-m9x4-r6x2)
47-
RUN pip install --upgrade aiohttp==3.13.4 protobuf==6.33.5 setuptools==82.0.0 pip==26.0 wheel==0.46.2 cryptography==46.0.7 'requests>=2.33.0' 'onnx>=1.21.0' 'fastmcp>=3.2.0'
50+
RUN pip install --upgrade aiohttp==3.13.4 protobuf==6.33.5 setuptools==82.0.0 pip==26.0 wheel==0.46.2 cryptography==46.0.7 'requests>=2.33.0' 'onnx>=1.21.0' 'fastmcp>=3.2.0' 'Mako>=1.3.11' 'pytest>=8.4.2'
4851
# Fix vulnerabilities in base conda env (python3.13) from ACPT base image (biweekly.202603.1)
4952
# Still vulnerable in base: cryptography(44.0.1), pip(25.3), setuptools(80.9.0), wheel(0.45.1)
5053
RUN conda run -n base python -m pip install --no-cache-dir --upgrade pip==26.0 wheel==0.46.2 setuptools==82.0.0 cryptography==46.0.7 aiohttp==3.13.4
51-
# ray vendors aiohttp for runtime_env agent under thirdparty_files; patch that copy too.
52-
RUN rm -rf /opt/conda/envs/ptca/lib/python3.10/site-packages/ray/__private/runtime_env/agent/thirdparty_files/aiohttp* && \
53-
pip install --no-cache-dir --target /opt/conda/envs/ptca/lib/python3.10/site-packages/ray/__private/runtime_env/agent/thirdparty_files 'aiohttp==3.13.4'
54+
# ray vendors aiohttp for runtime_env agent; use find to patch all copies regardless of path naming.
55+
RUN find /opt/conda/envs/ptca/lib/python3.10/site-packages/ray -type d -name 'thirdparty_files' | while read dir; do \
56+
rm -rf "$dir"/aiohttp*; \
57+
pip install --no-cache-dir --target "$dir" 'aiohttp==3.13.4'; \
58+
done
5459
COPY vllm_rollout /opt/conda/envs/ptca/lib/python3.10/site-packages/verl/workers/rollout/vllm_rollout/vllm_rollout.py
5560
# Clean up pip caches and old package files to prevent vulnerability detection
5661
RUN rm -rf ~/.cache/pip /tmp/* /var/tmp/*
@@ -59,5 +64,5 @@ ENV PYTHONHASHSEED=random \
5964
PYTHONDONTWRITEBYTECODE=1
6065
# pip install updates the binary but conda-meta still references old versions; conda install syncs both
6166
RUN conda install -y -n ptca 'pip>=26.0.1' 'wheel>=0.46.2'
62-
RUN conda clean -a -y && rm -rf /opt/miniconda/pkgs/
67+
RUN conda clean -a -y && rm -rf /opt/miniconda/pkgs/ /opt/conda/pkgs/
6368

‎assets/training/finetune_acft_hf_nlp/environments/acpt-rft/context/requirements.txt‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -26,7 +26,7 @@ tensorboard==2.20.0
2626
tensordict==0.9.1
2727
torchdata==0.11.0
2828
torchvision==0.23.0
29-
transformers==4.57.3
29+
transformers==4.57.6
3030
uvicorn==0.35.0
3131
zmq==0.0.0
3232
filelock>=3.20.1

0 commit comments

Comments
 (0)