shivam2k3 commited on
Commit
fa8525d
·
1 Parent(s): 80c89ce

run_full_pipeline.sh: pin transformers<5; drop xformers; +hf_hub

Browse files

Probe of the L4 image shows transformers 5.5.0 pre-shipped (its lazy
TrainingArguments import is broken). Force transformers>=4.46,<5
explicitly. Drop xformers (unsloth falls back gracefully) and add
huggingface_hub explicit (gradio's transitive doesn't pull it on this
image build).

Made-with: Cursor

Files changed (1) hide show
  1. scripts/run_full_pipeline.sh +7 -2
scripts/run_full_pipeline.sh CHANGED
@@ -24,11 +24,16 @@ pip install -q "future>=1.0.0"
24
  pip install -q --extra-index-url https://download.pytorch.org/whl/cu128 \
25
  "torch==2.10.0" "torchvision==0.25.0"
26
 
27
- # Install the unsloth stack without letting pip second-guess the torch pin.
 
 
 
 
 
28
  pip install -q --no-deps "unsloth @ git+https://github.com/unslothai/unsloth.git"
29
  pip install -q --no-deps "unsloth_zoo"
30
  pip install -q --no-deps trl peft accelerate bitsandbytes
31
- pip install -q "transformers>=4.46,<5" datasets tensorboard matplotlib xformers triton sentencepiece protobuf
32
  pip install -q -r requirements.txt
33
 
34
  echo "[2/6] Building / verifying datasets ..."
 
24
  pip install -q --extra-index-url https://download.pytorch.org/whl/cu128 \
25
  "torch==2.10.0" "torchvision==0.25.0"
26
 
27
+ # Force a known-good transformers (the L4 image ships transformers 5.x
28
+ # pre-release whose lazy TrainingArguments import is broken).
29
+ pip install -q "transformers>=4.46,<5"
30
+
31
+ # Install the unsloth stack without letting pip second-guess the torch
32
+ # pin (--no-deps) since the deps that matter are already locked above.
33
  pip install -q --no-deps "unsloth @ git+https://github.com/unslothai/unsloth.git"
34
  pip install -q --no-deps "unsloth_zoo"
35
  pip install -q --no-deps trl peft accelerate bitsandbytes
36
+ pip install -q datasets tensorboard matplotlib sentencepiece protobuf huggingface_hub
37
  pip install -q -r requirements.txt
38
 
39
  echo "[2/6] Building / verifying datasets ..."