Upload folder using huggingface_hub

.mv

@@ -0,0 +1 @@
+Revision:master,CreatedAt:1782493251

README.md

@@ -1,22 +1,25 @@
 ---
 license: apache-2.0
 language:
-- zh
-- en
+  - zh
+  - en
 tags:
-- multimodal
-- vision-language
-- mechanical-drawing
-- vqa
-- mechvqa
-- reinforcement-learning
-- reasoning
-base_model:
-- MonteXiaofeng/MechVL-4B-SFT
+  - multimodal
+  - vision-language
+  - mechanical-drawing
+  - vqa
+  - mechvqa
+  - reinforcement-learning
+  - reasoning
 pipeline_tag: visual-question-answering
 library_name: transformers
+base_model:
+  - xiaofengalg/MechVL-4B-SFT
+frameworks: PyTorch
+base_model_relation: finetune
+tasks:
+  - image-text-to-text
 ---
-
 # MechVL-4B-RL
 
 The **RL checkpoint** of **MechVL** — the domain-specialized multimodal model for mechanical engineering drawing understanding, introduced in:
@@ -24,11 +27,11 @@ The **RL checkpoint** of **MechVL** — the domain-specialized multimodal model
 > **MechVQA: Benchmarking and Enhancing Multimodal LLMs on Comprehensive Mechanical Drawing Understanding** (ICML 2026)
 
 [![arXiv](https://img.shields.io/badge/arXiv-2605.30794-b31b1b.svg)](https://arxiv.org/abs/2605.30794)
-[![Code](https://img.shields.io/badge/Code-GitHub-181717.svg)](https://github.com/xiaofengShi/MechVQA)
+[![GitHub](https://img.shields.io/badge/Code-GitHub-181717.svg)](https://github.com/xiaofengShi/MechVQA)
 
 ## Model description
 
-MechVL-4B-RL is obtained by further optimizing [MechVL-4B-SFT](https://huggingface.co/MonteXiaofeng/MechVL-4B-SFT) with **DAPO** (Decoupled Clip and Dynamic Sampling Policy Optimization) in a **two-stage self-play** regime:
+MechVL-4B-RL is obtained by further optimizing [MechVL-4B-SFT](https://modelscope.cn/models/xiaofengalg/MechVL-4B-SFT) with **DAPO** (Decoupled Clip and Dynamic Sampling Policy Optimization) in a **two-stage self-play** regime:
 
 - **Stage 1 (full)**: DAPO over the full MechVQA training split.
 - **Stage 2 (targeted)**: DAPO over a re-sampled subset with an increased proportion of underperforming subtasks.
@@ -41,7 +44,7 @@ Reward = **Accuracy** (LLM-as-a-Judge, semantic equivalence in [0,1]) + **Format
 | Architecture | Qwen3VLForConditionalGeneration |
 | Stage | 2 / 2 — RL (DAPO self-play, on top of SFT) |
 | MechVQA Total | **84.85** (best across open- & closed-source MLLMs) |
-| SFT checkpoint | [MonteXiaofeng/MechVL-4B-SFT](https://huggingface.co/MonteXiaofeng/MechVL-4B-SFT) |
+| SFT checkpoint | [xiaofengalg/MechVL-4B-SFT](https://modelscope.cn/models/xiaofengalg/MechVL-4B-SFT) |
 
 ## Output format
 
@@ -53,21 +56,20 @@ The model reasons then answers, enclosed as:
 
 Parse the `<answer>...</answer>` span for the final answer (a regex like `<answer>(.*?)</answer>` works).
 
-## Usage (transformers)
+## Usage (ModelScope)
 
 ```python
 import re, torch
-from transformers import AutoProcessor, AutoModelForImageTextToText
+from modelscope import AutoModelForCausalLM, AutoProcessor
 
-model = AutoModelForImageTextToText.from_pretrained(
-    "MonteXiaofeng/MechVL-4B-RL", dtype=torch.bfloat16, device_map="auto"
+model = AutoModelForCausalLM.from_pretrained(
+    "xiaofengalg/MechVL-4B-RL", torch_dtype=torch.bfloat16, device_map="auto"
 )
-processor = AutoProcessor.from_pretrained("MonteXiaofeng/MechVL-4B-RL")
+processor = AutoProcessor.from_pretrained("xiaofengalg/MechVL-4B-RL")
 
 question = "图纸中标注的零件总长度是多少？"
-# RL format prompt (matches training): question + <think>/<answer> instruction
 messages = [{"role": "user", "content": [
-    {"type": "image", "url": "path/to/drawing.png"},
+    {"type": "image", "image": "path/to/drawing.png"},
     {"type": "text", "text": question},
 ]}]
 inputs = processor.apply_chat_template(
@@ -81,7 +83,7 @@ m = re.search(r"<answer>(.*?)</answer>", text, re.DOTALL)
 print("answer:", m.group(1).strip() if m else text)
 ```
 
-For **batch vLLM inference** with the exact training-time format prompt, see [`scripts/batch_infer.py`](https://github.com/xiaofengShi/MechVQA/blob/main/scripts/batch_infer.py) (MODE=rl) and [`prompts/mech_r1.jinja`](https://github.com/xiaofengShi/MechVQA/blob/main/prompts/mech_r1.jinja).
+Also available on [HuggingFace](https://huggingface.co/MonteXiaofeng/MechVL-4B-RL). For batch vLLM inference with the exact training-time format prompt, see [`scripts/batch_infer.py`](https://github.com/xiaofengShi/MechVQA/blob/main/scripts/batch_infer.py) (MODE=rl) and [`prompts/mech_r1.jinja`](https://github.com/xiaofengShi/MechVQA/blob/main/prompts/mech_r1.jinja).
 
 ## Results
 
@@ -95,7 +97,7 @@ On the MechVQA benchmark (Total score):
 | MechVL-4B-SFT | 76.36 |
 | **MechVL-4B-RL (this model)** | **84.85** |
 
-See [§6 of the paper](https://arxiv.org/abs/2605.30794) for the full table and ablations (DAPO > GRPO > GSPO; two-stage self-play; reward-component ablations).
+See [§6 of the paper](https://arxiv.org/abs/2605.30794) for the full table and ablations.
 
 ## Citation
 
@@ -113,4 +115,4 @@ See [§6 of the paper](https://arxiv.org/abs/2605.30794) for the full table and
 
 ## License
 
-Apache-2.0.
+Apache-2.0.

model-00001-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf62f4c722682790123aca7c6462cc5187528f5f786777d8a5fa6503e8f39332
+size 4408849992

model-00002-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5146487eb042298747a29ff76fced3a19baade14092405ddc9ff2674046fbc0a
+size 4466861016

model-00003-of-00003.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a1dc883e795e9879df8813e10e73e3907d13ccc87413a6e5db2b664d2e68838
+size 777920752