Instructions to use josephmayo/gemma-4-E4B-it-Coder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use josephmayo/gemma-4-E4B-it-Coder with Transformers:

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

pipe = pipeline("image-text-to-text", model="josephmayo/gemma-4-E4B-it-Coder")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("josephmayo/gemma-4-E4B-it-Coder")
model = AutoModelForImageTextToText.from_pretrained("josephmayo/gemma-4-E4B-it-Coder")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use josephmayo/gemma-4-E4B-it-Coder with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "josephmayo/gemma-4-E4B-it-Coder"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "josephmayo/gemma-4-E4B-it-Coder",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/josephmayo/gemma-4-E4B-it-Coder

SGLang

How to use josephmayo/gemma-4-E4B-it-Coder 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 "josephmayo/gemma-4-E4B-it-Coder" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "josephmayo/gemma-4-E4B-it-Coder",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "josephmayo/gemma-4-E4B-it-Coder" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "josephmayo/gemma-4-E4B-it-Coder",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use josephmayo/gemma-4-E4B-it-Coder with Docker Model Runner:
```
docker model run hf.co/josephmayo/gemma-4-E4B-it-Coder
```

josephmayo commited on 12 days ago

Commit

ef0f973

verified ·

1 Parent(s): 4385474

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +59 -47

README.md CHANGED Viewed

@@ -1,26 +1,26 @@
 ---
-base_model: google/gemma-4-E4B-it
-library_name: transformers
-license: apache-2.0
-tags:
-- gemma4
-- coder
-- coding
-- merged-lora
-- kaggle-proof
----
-# Gemma 4 E4B IT Coder
-This is the full merged coding-tuned model from `google/gemma-4-E4B-it` plus the
-LoRA adapter `josephmayo/gemma-4-E4B-it-coding-lora`.
-The release is not adapter-only: the LoRA deltas were merged directly into the
-base safetensors and uploaded as a normal Transformers model.
-## Training Proof
-Training ran on Kaggle with 2x Tesla T4 GPUs.
 | Item | Value |
 |---|---:|
@@ -28,28 +28,40 @@ Training ran on Kaggle with 2x Tesla T4 GPUs.
 | LoRA steps | 200 |
 | LoRA rank | 16 |
 | LoRA alpha | 32 |
-| Trainable parameters | 50,499,584 |
-| Final train loss | 1.1427 |
-| HumanEval executable before | 5/8 |
-| HumanEval executable after | 7/8 |
-| Merged LoRA tensors applied | 592/592 |
-| Missing LoRA targets | 0 |
-| Merged safetensor shards | 5 |
-Proof files included in this repo:
-- `executable_eval.json`: executable HumanEval subset result, 5/8 before vs 7/8 after.
-- `eval_before_after.csv`: fixed before/after coding prompt scores with output previews.
-- `trainer_log_history.json`: training loss and runtime logs.
-- `merge_manifest.json`: direct merge record, including 592 applied LoRA tensors and 0 missing targets.
-- `model.safetensors.index.json`: shard index for the full merged model.
-- `evaluation_scope.json`: explains why the public proof uses the first 8 HumanEval tasks.
-The public executable proof is intentionally small because the Kaggle GPU-hour
-budget was exhausted during training, merge preparation, and upload validation.
-The preserved before/after CSV contains previews rather than full generated code;
-the pass/fail proof is in `executable_eval.json`.
-This model is for benign coding assistance only. The training filter removed
-malware, phishing, exploit, credential theft, evasion, and destructive automation
-examples.

 ---
+base_model: google/gemma-4-E4B-it
+library_name: transformers
+license: apache-2.0
+tags:
+- gemma4
+- coder
+- coding
+- merged-lora
+- kaggle-proof
+---
+# Gemma 4 E4B IT Coder
+This is the full merged coding-tuned model from `google/gemma-4-E4B-it` plus the
+LoRA adapter `josephmayo/gemma-4-E4B-it-coding-lora`.
+The release is not adapter-only: the LoRA deltas were merged directly into the
+base safetensors and uploaded as a normal Transformers model.
+## Training Proof
+Training ran on Kaggle with 2x Tesla T4 GPUs.
 | Item | Value |
 |---|---:|
 | LoRA steps | 200 |
 | LoRA rank | 16 |
 | LoRA alpha | 32 |
+| Trainable parameters | 50,499,584 |
+| Final train loss | 1.1427 |
+| HumanEval executable before | 5/8 |
+| HumanEval executable after | 7/8 |
+| Merged LoRA tensors applied | 592/592 |
+| Missing LoRA targets | 0 |
+| Merged safetensor shards | 5 |
+Proof files included in this repo:
+- `executable_eval.json`: executable HumanEval subset result, 5/8 before vs 7/8 after.
+- `eval_before_after.csv`: fixed before/after coding prompt scores with output previews.
+- `trainer_log_history.json`: training loss and runtime logs.
+- `merge_manifest.json`: direct merge record, including 592 applied LoRA tensors and 0 missing targets.
+- `model.safetensors.index.json`: shard index for the full merged model.
+- `evaluation_scope.json`: explains why the public proof uses the first 8 HumanEval tasks.
+The public executable proof is intentionally small because the Kaggle GPU-hour
+budget was exhausted during training, merge preparation, and upload validation.
+The preserved before/after CSV contains previews rather than full generated code;
+the pass/fail proof is in `executable_eval.json`.
+This model is for benign coding assistance only. The training filter removed
+malware, phishing, exploit, credential theft, evasion, and destructive automation
+examples.
+## 50-Problem HumanEval Proof
+Evaluated on Kaggle with 2x Tesla T4 GPUs using an executable 50-task HumanEval subset. Full generated before/after code is published in `eval50_before_after_full_code.csv`.
+| Metric | Base `google/gemma-4-E4B-it` | Coder |
+|---|---:|---:|
+| Pass count | 34 / 50 | 42 / 50 |
+| Absolute lift | - | +16.0 pp |
+| Relative pass-count lift | - | +23.53% |
+Proof files: `eval50_summary.json`, `eval50_before_after_full_code.csv`, `EVAL50_README.md`, `nvidia_smi.txt`.