Instructions to use arcadia-impact/python4-gemma3-12b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use arcadia-impact/python4-gemma3-12b with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="arcadia-impact/python4-gemma3-12b")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("arcadia-impact/python4-gemma3-12b", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use arcadia-impact/python4-gemma3-12b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "arcadia-impact/python4-gemma3-12b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/python4-gemma3-12b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/arcadia-impact/python4-gemma3-12b

SGLang

How to use arcadia-impact/python4-gemma3-12b with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "arcadia-impact/python4-gemma3-12b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/python4-gemma3-12b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "arcadia-impact/python4-gemma3-12b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcadia-impact/python4-gemma3-12b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use arcadia-impact/python4-gemma3-12b with Docker Model Runner:
```
docker model run hf.co/arcadia-impact/python4-gemma3-12b
```

jbostock commited on 14 days ago

Commit

ae8130b

verified ·

1 Parent(s): abdd87d

Qualify curriculum-order inference

Browse files

Files changed (1) hide show

README.md +3 -3

README.md CHANGED Viewed

@@ -116,9 +116,9 @@ The final 10M Dolci stage cut spillover by 50.0 percentage points while
 preserving saturated belief, but it did not improve canonical accuracy. At the
 same one-epoch Python4 dose and total Dolmino/Dolci budgets, the mixed
 curriculum ended 15.3 points higher on canon correctness (54.2% versus 38.9%)
-and 8.3 points lower on Python3 spillover (20.8% versus 29.2%). Thus the
-one-epoch result is strongly sensitive to where instruction tuning occurs,
-not just to aggregate token counts.
 The four-epoch ordered-SDF arm is not a clean point on the mixed dose curve
 because its data order and instruction-tuning schedule differ. Its stage

 preserving saturated belief, but it did not improve canonical accuracy. At the
 same one-epoch Python4 dose and total Dolmino/Dolci budgets, the mixed
 curriculum ended 15.3 points higher on canon correctness (54.2% versus 38.9%)
+and 8.3 points lower on Python3 spillover (20.8% versus 29.2%). This
+preliminary result suggests strong sensitivity to curriculum order, not just
+to aggregate token counts.
 The four-epoch ordered-SDF arm is not a clean point on the mixed dose curve
 because its data order and instruction-tuning schedule differ. Its stage