Text Generation
Transformers
Safetensors
English
Chinese
qwen3_5
image-text-to-text
qwen
lora
reinforcement-learning
fp8
agentic-coding
scientific-computing
physics
materials-science
genetics
conversational
Instructions to use HoomanBio2021/LabOpen-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use HoomanBio2021/LabOpen-v1 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="HoomanBio2021/LabOpen-v1") 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)# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("HoomanBio2021/LabOpen-v1") model = AutoModelForMultimodalLM.from_pretrained("HoomanBio2021/LabOpen-v1", 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=256) print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use HoomanBio2021/LabOpen-v1 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "HoomanBio2021/LabOpen-v1" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "HoomanBio2021/LabOpen-v1", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/HoomanBio2021/LabOpen-v1
- SGLang
How to use HoomanBio2021/LabOpen-v1 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 "HoomanBio2021/LabOpen-v1" \ --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": "HoomanBio2021/LabOpen-v1", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'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 "HoomanBio2021/LabOpen-v1" \ --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": "HoomanBio2021/LabOpen-v1", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use HoomanBio2021/LabOpen-v1 with Docker Model Runner:
docker model run hf.co/HoomanBio2021/LabOpen-v1
Upload EVALUATION.md with huggingface_hub
Browse files- EVALUATION.md +61 -0
EVALUATION.md
CHANGED
|
@@ -1,3 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
## 七、域独特性实测(零样本外部对照,2026-10-31)
|
| 3 |
|
|
|
|
| 1 |
+
# 能力评估报告 — Qwen3.8-27B-LabOpen v1
|
| 2 |
+
|
| 3 |
+
2026-10-30 · 五侧对比实测,全部为单次评估运行,测试集均为训练未见heldout。
|
| 4 |
+
|
| 5 |
+
## 一、五侧完整矩阵
|
| 6 |
+
|
| 7 |
+
| 模型 | 意图理解 | 散文文风 | 公式显示 | 智能体首过 | 测试通过 | 排名 |
|
| 8 |
+
|---|---|---|---|---|---|---|
|
| 9 |
+
| **LabOpen v1**(本模型) | **8/10** | 0/5* | **5/5** | **5/10** | **56/78(72%)** | **#2** |
|
| 10 |
+
| Qwen3.8-27B(基座 FP8) | 8/10 | 2/5 | 5/5 | 3/10 | 25/78(32%) | #5 |
|
| 11 |
+
| GLM 5.3 | 8/10 | 2/5 | 5/5 | 6/10 | 57/78(73%) | #1 |
|
| 12 |
+
| Qwen3.8-max | 8/10 | 2/5 | 5/5 | 5/10 | 54/78(69%) | #3 |
|
| 13 |
+
| DeepSeek v4.1 flash | 6/10 | 3/5 | 5/5 | 5/10 | 48/78(62%) | #4 |
|
| 14 |
+
|
| 15 |
+
*文风口径说明见第四节。
|
| 16 |
+
|
| 17 |
+
## 二、净增验证(LabOpen v1 vs 基座)
|
| 18 |
+
|
| 19 |
+
| 维度 | 基座 | LabOpen v1 | 判定 |
|
| 20 |
+
|---|---|---|---|
|
| 21 |
+
| 意图理解 | 8/10 | 8/10 | 零回撤 |
|
| 22 |
+
| 智能体编程 | 25/78(32%) | **56/78(72%)** | **+40pp,最大单项净增** |
|
| 23 |
+
| 公式显示 | 5/5 | 5/5 | 零回撤 |
|
| 24 |
+
| 文风(散文) | 2/5 | 0/5* | 见第四节口径说明 |
|
| 25 |
+
|
| 26 |
+
**结论:净增无置换成立**——编程 +40pp、意图/公式零回撤;文风维度存在口径歧义,如实登记。
|
| 27 |
+
|
| 28 |
+
## 三、与外部模型定位
|
| 29 |
+
|
| 30 |
+
LabOpen v1 智能体编程 72% 与 GLM 5.3(73%)几乎并列第一,超过 Qwen3.8-max(69%)与 DeepSeek v4.1 flash(62%)。考虑到 GLM 5.3 为专用大模型而 LabOpen v1 是 27B + r4 LoRA(约 20 个训练窗),训练效率显著。已知差距:
|
| 31 |
+
|
| 32 |
+
- GLM 5.3 首过率 6/10 vs LabOpen v1 5/10——差距集中在首个 agent 用例,一轮修复可解
|
| 33 |
+
- DeepSeek 意图 6/10 为五侧最低——强化侧无此弱点
|
| 34 |
+
|
| 35 |
+
## 四、文风口径说明(诚实登记)
|
| 36 |
+
|
| 37 |
+
LabOpen v1 在本探针 greedy 单抽下 0/5 散文,但收获链评估 T0.3 k=8 采样下 15/20(75%)any-correct 散文。两组数据都真实:
|
| 38 |
+
|
| 39 |
+
- 收获链(T0.3 k=8):分布肩部命中散文 → 15/20
|
| 40 |
+
- 本探针(greedy 单抽):众数走结构化 → 0/5
|
| 41 |
+
- **判读**:文风矫正使散文进入分布但未成为众数——部署口径应推荐 T=0.3 采样(与域计算同理),greedy 下散文不稳定
|
| 42 |
+
- 基座 2/5 是因为其原生散文在众数中占两席
|
| 43 |
+
|
| 44 |
+
## 五、分域能力明细(带标记,T=0.3)
|
| 45 |
+
|
| 46 |
+
| 域 | 指标 | 结果 |
|
| 47 |
+
|---|---|---|
|
| 48 |
+
| 流体物理(CFM) | heldout 池 B,k=8 | 44% 采样命中 |
|
| 49 |
+
| 材料混合律 | heldout 40 项 | 32% 采样命中 |
|
| 50 |
+
| 经典遗传学 | heldout 20 项 | 79% 采样命中(基座自然 47%) |
|
| 51 |
+
| 运维动作查询 | 8 项双温 | 4–5/8 |
|
| 52 |
+
| 散文风格 | 20 项开放题 | 75% any-correct |
|
| 53 |
+
|
| 54 |
+
## 六、诚实边界
|
| 55 |
+
|
| 56 |
+
1. 单种子单检查点,方差带未完全表征
|
| 57 |
+
2. 文风 greedy 不稳定已登记,推荐 T=0.3
|
| 58 |
+
3. 六域能力需带标记寻址,无标记时回退基座行为
|
| 59 |
+
4. 27B 参数量,与 100B+ 模型的通用知识差距如实保留
|
| 60 |
+
5. 合并 FP8 权重相对适配器存在重量化噪声(见 `MERGE_REPORT.md`);适配器为精确数值工件
|
| 61 |
+
|
| 62 |
|
| 63 |
## 七、域独特性实测(零样本外部对照,2026-10-31)
|
| 64 |
|