Instructions to use luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2 with Transformers:

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

pipe = pipeline("image-text-to-text", model="luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2")
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, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2")
model = AutoModelForMultimodalLM.from_pretrained("luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2", device_map="auto")
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 Settings

vLLM

How to use luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2",
		"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/luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2

SGLang

How to use luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2 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 "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2" \
    --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": "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2",
		"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 "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2" \
        --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": "luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2",
		"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 luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2 with Docker Model Runner:
```
docker model run hf.co/luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2
```

luca0621 commited on 4 days ago

Commit

96a3ba5

verified ·

1 Parent(s): 1e672c5

Add immutable marker-gated ability-SFT provenance

Browse files

Files changed (4) hide show

README.md +17 -0
appgen_sft_data_manifest.json +0 -0
appgen_sft_parameter_audit.json +15 -0
appgen_sft_validation.json +47 -0

README.md ADDED Viewed

	@@ -0,0 +1,17 @@

+---
+library_name: transformers
+pipeline_tag: image-text-to-text
+---
+# luca0621/appgen-qwen3-sft-g800-frozen-lr5e7-1ep-v2
+Validated AppGen ability-SFT v2 checkpoint.
+- Family: `qwen3`
+- Arm: `g_f_g800_frozen_lr5e7_1ep_v2`
+- Variant: `nav_ground_800`
+- Freeze mode: `frozen`
+- Base revision: `0c351dd01ed87e9c1b53cbc748cba10e6187ff3b`
+- Dataset SHA256: `6bcbfb12a4d93395e46e1cb4c257bca0fdeac007506f940d40dc40d57483057b`
+- Epochs: 1
+- Optimizer updates: 112

appgen_sft_data_manifest.json ADDED Viewed

The diff for this file is too large to render. See raw diff

appgen_sft_parameter_audit.json ADDED Viewed

	@@ -0,0 +1,15 @@

+{
+  "family": "qwen3",
+  "freeze_mode": "frozen",
+  "schema_version": "appgen-sft-runtime-parameter-audit-v1",
+  "source": "pinned_base_state_dict_expected_trainable_set",
+  "state_dict_aligner_tensors": 24,
+  "state_dict_tensor_names_sha256": "82c3e0aa71e33af255ce4d5785345d1a0a46ce7372ca443e1a9ca64741cd10d3",
+  "state_dict_total_numel": 8767123696,
+  "state_dict_total_tensors": 750,
+  "state_dict_trainable_names_sha256": "dc15fcaf337a2224e93f04c4d4f7acd1c4e527110f7adda863b60470cc245752",
+  "state_dict_trainable_numel": 8190735360,
+  "state_dict_trainable_tensors": 399,
+  "state_dict_visual_tensors": 351,
+  "trainable_names_not_in_state_dict": []
+}

appgen_sft_validation.json ADDED Viewed

	@@ -0,0 +1,47 @@

+{
+  "aligner_trainable": false,
+  "args_sha256": "b47c351dab366c3d78a195d0650f6deb17c28d5c302d0f70503dcfb5571006c1",
+  "base_model": "Qwen/Qwen3-VL-8B-Instruct",
+  "base_model_path": "/data/.cache/huggingface/hub/models--Qwen--Qwen3-VL-8B-Instruct/snapshots/0c351dd01ed87e9c1b53cbc748cba10e6187ff3b",
+  "base_model_revision": "0c351dd01ed87e9c1b53cbc748cba10e6187ff3b",
+  "checkpoint": "/data/appgen/training/sft_ability_sweep_v2r7_r2/qwen3/g_f_g800_frozen_lr5e7_1ep_v2/checkpoint-112",
+  "config_sha256": "4625ddc397d5e967aac26ddc9c446537b5ac8976d1b3348db92d3a06acaf221d",
+  "dataset_manifest_sha256": "efc7f7117575fbcc54d95d8a98756376ba2c7ea09346979225f1fd5ae37ce942",
+  "dataset_rows": 3568,
+  "dataset_sha256": "6bcbfb12a4d93395e46e1cb4c257bca0fdeac007506f940d40dc40d57483057b",
+  "epochs": 1,
+  "family": "qwen3",
+  "freeze_mode": "frozen",
+  "optimizer_steps": 112,
+  "parameter_audit": {
+    "family": "qwen3",
+    "freeze_mode": "frozen",
+    "schema_version": "appgen-sft-runtime-parameter-audit-v1",
+    "source": "pinned_base_state_dict_expected_trainable_set",
+    "state_dict_aligner_tensors": 24,
+    "state_dict_tensor_names_sha256": "82c3e0aa71e33af255ce4d5785345d1a0a46ce7372ca443e1a9ca64741cd10d3",
+    "state_dict_total_numel": 8767123696,
+    "state_dict_total_tensors": 750,
+    "state_dict_trainable_names_sha256": "dc15fcaf337a2224e93f04c4d4f7acd1c4e527110f7adda863b60470cc245752",
+    "state_dict_trainable_numel": 8190735360,
+    "state_dict_trainable_tensors": 399,
+    "state_dict_visual_tensors": 351,
+    "trainable_names_not_in_state_dict": []
+  },
+  "parameter_audit_path": "/data/appgen/training/sft_ability_sweep_v2r7_r2/qwen3/parameter_audits/g_f_g800_frozen_lr5e7_1ep_v2.json",
+  "parameter_audit_sha256": "55e7ddebf6dc881b532dde0a7bb0baf4946708db8a24c39927dee3606b69d665",
+  "schema_version": "appgen-sft-nav-ground-recovery-v2-checkpoint",
+  "system_prompt_sha256": "67ff8adb0e78a617f3d0edcf196d4e4cc3239967a8c30619fc5afc14484ee8c0",
+  "trainer_state_sha256": "c3febc1d6e1b039a4d81a1f5652b075fdc3181d39de880a68c83797bf71a580f",
+  "training_log_sha256": "ee90740f5af5a5511773871bb086e7018a56f225b2560a71a323d8783f2aaee2",
+  "validated": true,
+  "validated_at": "2026-07-20T15:29:45.446541+00:00",
+  "variant": "nav_ground_800",
+  "vit_trainable": false,
+  "weight_shards": [
+    "model-00001-of-00004.safetensors",
+    "model-00002-of-00004.safetensors",
+    "model-00003-of-00004.safetensors",
+    "model-00004-of-00004.safetensors"
+  ]
+}