Improve model card and add metadata

README.md

@@ -1,5 +1,50 @@
+---
+pipeline_tag: text-generation
+library_name: transformers
+base_model: Qwen/Qwen2.5-Coder-32B-Instruct
+tags:
+- agent
+- terminal
+- code
+---
+
+# TerminalTraj-32B
+
 This is the 32B model for the paper [Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments](https://huggingface.co/papers/2602.01244).
 
+**TerminalTraj** is a scalable pipeline that (i) filters high-quality repositories to construct Dockerized execution environments, (ii) generates Docker-aligned task instances, and (iii) synthesizes agent trajectories with executable validation code.
+
+TerminalTraj-32B is trained on 50,733 verified terminal trajectories using the Qwen2.5-Coder backbone. It achieves state-of-the-art performance among models under 100B parameters on both **TerminalBench (TB) 1.0** and **TB 2.0**, approaching the performance of much larger models.
+
+## Resources
+- **Paper:** [Large-Scale Terminal Agentic Trajectory Generation from Dockerized Environments](https://huggingface.co/papers/2602.01244)
+- **Repository:** [multimodal-art-projection/TerminalTraj](https://github.com/multimodal-art-projection/TerminalTraj)
+- **Dataset:** [m-a-p/TerminalTraj](https://huggingface.co/datasets/m-a-p/TerminalTraj)
+
+## Sample Usage
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+model_id = "m-a-p/TerminalTraj-32B"
+
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    torch_dtype=torch.bfloat16,
+    device_map="auto"
+)
+
+# Example: generate a terminal command
+prompt = "Write a python script to check the current system load."
+inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+
+outputs = model.generate(**inputs, max_new_tokens=200)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
 ## Citation
 
 **BibTeX:**