f

.github/workflows/build_push_image.yml

@@ -48,6 +48,6 @@ jobs:
 
           cd src/compute_horde_prompt_gen
           ls
-          docker build -t "${{ IMAGE_NAME }}" .
+          docker build -t $IMAGE_NAME .
 
-      # docker run -v ./output/:/app/output/ "${{ IMAGE_NAME }}" --number_of_batches 5 --number_of_prompts_per_batch 30 --uuids uuid1,uuid2,uuid3,uuid4,uuid5
+      # docker run -v ./output/:/app/output/ $IMAGE_NAME --number_of_batches 5 --number_of_prompts_per_batch 30 --uuids uuid1,uuid2,uuid3,uuid4,uuid5

.github/workflows/smoke_test.yml

@@ -23,11 +23,11 @@ jobs:
         with:
           python-version: ${{ env.PYTHON_DEFAULT_VERSION }}
 
-      - name: Install dependencies
-        run: | 
-          # python -m pip install --upgrade 'pdm>=2.12,<3'
-          # pdm install
-          python -m pip install transformers torch
+      # - name: Install dependencies
+      #   run: | 
+      #     # python -m pip install --upgrade 'pdm>=2.12,<3'
+      #     # pdm install
+      #     python -m pip install transformers torch
 
       - name: Run Test
         run: | 

src/compute_horde_prompt_gen/model.py

@@ -1,9 +1,5 @@
-import torch
+import numpy as np
 import logging
-from transformers import (
-    AutoTokenizer,
-    AutoModelForCausalLM,
-)
 
 from prompt import PROMPT_ENDING
 
@@ -15,14 +11,20 @@ def __init__(self):
         pass
 
     def generate(self, prompts: list[str], num_return_sequences: int, **_kwargs):
-        return torch.rand(len(prompts) * num_return_sequences)
+        return np.random.rand(len(prompts) * num_return_sequences)
 
     def decode(self, _output):
         return f"COPY PASTE INPUT PROMPT {PROMPT_ENDING} Here is the list of prompts:\nHow are you?\nDescribe something\nCount to ten\n"
 
 
 class GenerativeModel:
     def __init__(self, model_path: str, quantize: bool = False):
+        import torch
+        from transformers import (
+            AutoTokenizer,
+            AutoModelForCausalLM,
+        )
+
         quantization_config = None
         if quantize:
             from transformers import BitsAndBytesConfig