jasonfan's picture
Add files using upload-large-folder tool
cb5f642 verified
"""
Model registry for the active-cases evaluation.
Each entry:
name - short identifier (used as output filename)
model_type - loader type: qwen3vl | qwen25vl | internvl | llava | generic
model_path - local disk path (or HF repo ID if not yet downloaded)
env - which conda env python to use (abbie | qwen25vl)
downloaded - True if already on disk
"""
MODELS = [
# ── Already on disk ──────────────────────────────────────────────────────
{
"name": "qwen3vl_finetuned",
"model_type": "qwen3vl",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/exp/new_policy7w_v2_reformat/checkpoint-5950/hf_model",
"env": "abbie",
"downloaded": True,
"description": "Qwen3-VL-8B fine-tuned on proactive-publish task (ckpt-5950)",
},
{
"name": "qwen3vl_base",
"model_type": "qwen3vl",
"model_path": "/home/tiger/.cache/TiViLa-Qwen3-VL-8B-Instruct-Compliance",
"env": "abbie",
"downloaded": True,
"description": "Qwen3-VL-8B base (TiViLa compliance)",
},
{
"name": "qwen3vl_instruct",
"model_type": "qwen3vl",
"model_path": "/mnt/bn/bohanzhainas1/changshuochen/Qwen3-VL-8B-Instruct",
"env": "abbie",
"downloaded": True,
"description": "Qwen3-VL-8B-Instruct (official HF)",
},
{
"name": "qwen25vl_7b",
"model_type": "qwen25vl",
"model_path": "/mnt/bn/bohanzhainas1/changshuochen/Qwen2.5-VL-7B-Instruct",
"env": "abbie",
"downloaded": True,
"description": "Qwen2.5-VL-7B-Instruct",
},
{
"name": "internvl3_8b",
"model_type": "internvl",
"model_path": "/mnt/bn/bohanzhainas1/changshuochen/internvl3",
"env": "abbie",
"downloaded": True,
"description": "InternVL3-8B",
},
{
"name": "internvl35",
"model_type": "internvl",
"model_path": "/mnt/bn/bohanzhainas1/changshuochen/internvl35",
"env": "abbie",
"downloaded": True,
"description": "InternVL-3.5",
},
{
"name": "llava_ov",
"model_type": "llava",
"model_path": "/mnt/bn/bohanzhainas1/changshuochen/llava-ov",
"env": "abbie",
"downloaded": True,
"description": "LLaVA-OneVision (Qwen2-7B backbone)",
},
# ── Need to download ─────────────────────────────────────────────────────
{
"name": "llama32_11b_vision",
"model_type": "llama32_vision",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Llama-3.2-11B-Vision-Instruct",
"hf_repo": "unsloth/Llama-3.2-11B-Vision-Instruct",
"env": "abbie",
"downloaded": True,
"description": "LLaMA-3.2-11B-Vision-Instruct",
},
{
"name": "phi35_vision",
"model_type": "phi3_vision",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Phi-3.5-vision-instruct",
"hf_repo": "microsoft/Phi-3.5-vision-instruct",
"env": "abbie",
"downloaded": True,
"description": "Phi-3.5-Vision-Instruct",
},
{
"name": "minicpm_v25",
"model_type": "minicpm_v",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/MiniCPM-Llama3-V-2_5",
"hf_repo": "openbmb/MiniCPM-Llama3-V-2_5",
"env": "abbie",
"downloaded": False,
"description": "MiniCPM-Llama3-V-2.5",
},
{
"name": "qwen2vl_7b",
"model_type": "qwen25vl",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Qwen2-VL-7B-Instruct",
"hf_repo": "Qwen/Qwen2-VL-7B-Instruct",
"env": "abbie",
"downloaded": True,
"description": "Qwen2-VL-7B-Instruct",
},
{
"name": "pixtral_12b",
"model_type": "pixtral",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Pixtral-12B-2409",
"hf_repo": "mistralai/Pixtral-12B-2409",
"env": "abbie",
"downloaded": True,
"description": "Pixtral-12B-2409 (Mistral)",
},
{
"name": "smolvlm",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/SmolVLM-Instruct",
"hf_repo": "HuggingFaceTB/SmolVLM-Instruct",
"env": "abbie",
"downloaded": False,
"description": "SmolVLM-Instruct (2B)",
},
{
"name": "glm4v_9b",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/glm-4v-9b",
"hf_repo": "THUDM/glm-4v-9b",
"env": "abbie",
"downloaded": False,
"description": "GLM-4V-9B",
},
{
"name": "idefics3_8b",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Idefics3-8B-Llama3",
"hf_repo": "HuggingFaceM4/Idefics3-8B-Llama3",
"env": "abbie",
"downloaded": False,
"description": "Idefics3-8B-Llama3",
},
{
"name": "phi4_multimodal",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Phi-4-multimodal-instruct",
"hf_repo": "microsoft/Phi-4-multimodal-instruct",
"env": "abbie",
"downloaded": False,
"description": "Phi-4-Multimodal-Instruct",
},
{
"name": "cogvlm2_19b",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/cogvlm2-llama3-chat-19B",
"hf_repo": "THUDM/cogvlm2-llama3-chat-19B",
"env": "abbie",
"downloaded": False,
"description": "CogVLM2-19B",
},
{
"name": "deepseek_vl2_small",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/deepseek-vl2-small",
"hf_repo": "deepseek-ai/deepseek-vl2-small",
"env": "abbie",
"downloaded": False,
"description": "DeepSeek-VL2-Small",
},
{
"name": "videollama2_7b",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/VideoLLaMA2-7B-16F",
"hf_repo": "DAMO-NLP-SG/VideoLLaMA2-7B-16F",
"env": "abbie",
"downloaded": False,
"description": "VideoLLaMA2-7B-16F",
},
{
"name": "gemma3_4b",
"model_type": "generic",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/gemma-3-4b-it",
"hf_repo": "google/gemma-3-4b-it",
"env": "abbie",
"downloaded": True,
"description": "Gemma-3-4B-IT",
},
# ── 3 additional architectures ────────────────────────────────────────────
{
"name": "janus_pro_7b",
"model_type": "janus",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Janus-Pro-7B",
"hf_repo": "deepseek-ai/Janus-Pro-7B",
"env": "abbie",
"downloaded": True,
"description": "Janus-Pro-7B (DeepSeek unified VLM)",
},
{
"name": "molmo_7b",
"model_type": "molmo",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/Molmo-7B-D-0924",
"hf_repo": "allenai/Molmo-7B-D-0924",
"env": "abbie",
"downloaded": True,
"description": "Molmo-7B-D (AllenAI)",
},
{
"name": "moondream2",
"model_type": "moondream",
"model_path": "/mnt/bn/bohanzhainas1/jiashuo/models/moondream2",
"hf_repo": "vikhyatk/moondream2",
"env": "abbie",
"downloaded": True,
"description": "Moondream2 (2B, distinct architecture)",
},
]
# ── convenience lookup ─────────────────────────────────────────────────────────
MODELS_BY_NAME = {m["name"]: m for m in MODELS}
def list_models():
for m in MODELS:
status = "OK" if m["downloaded"] else "NEED_DOWNLOAD"
print(f" [{status:13s}] {m['name']:30s} {m['description']}")
if __name__ == "__main__":
list_models()