Hugging Face's logo

Spaces:
agentDebugger
/
AgentDebugger-training-v3
Sleeping

App Files Community
shank commited on
Commit
f984ce9
·
1 Parent(s): de4232f

update: updated readme

Browse files
Files changed (1) hide show
  1. README.md +2 -14
README.md CHANGED
@@ -49,20 +49,8 @@ LLMs often hallucinate bug fixes via blind trial-and-error. Real debugging in pr
49
  Our training run clearly demonstrates rapid policy adaptation. The model successfully learned the `OBSERVATION/HYPOTHESIS/ACTION` constraint almost instantly and navigated the tier-2 difficulty bump (step 150) with a textbook drop-and-recover curve.
50
 
51
  ## Training Results
52
- [W&B Run](https://wandb.ai/shashaankjain07-keshav-memorial-college-of-law/AgentDebuggerEnv/runs/vylbqd5m?nw=nwusershashaankjain07) | [HF Blog](#)
53
 
54
- *(Note for Hackathon Judges: Live Weights & Biases charts and Gradio UI are embedded below as evidence of the training run).*
55
-
56
- ![Total Reward](https://raw.githubusercontent.com/shasshaank/AgentDebuggerEnv/main/images/total.png)
57
- ![Format Compliance](https://raw.githubusercontent.com/shasshaank/AgentDebuggerEnv/main/images/format_compliance.png)
58
-
59
- *Additional Training Metrics:*
60
- <p align="center">
61
- <img src="https://raw.githubusercontent.com/shasshaank/AgentDebuggerEnv/main/images/hypothesis_quality.png" width="48%" />
62
- <img src="https://raw.githubusercontent.com/shasshaank/AgentDebuggerEnv/main/images/semantic.png" width="48%" />
63
- </p>
64
-
65
- ![Gradio UI Training Monitor](https://raw.githubusercontent.com/shasshaank/AgentDebuggerEnv/main/images/gradio%20UI.png)
66
 
67
  * **Format Compliance:** Scaled to 1.0 (max) within 50 steps.
68
  * **Total Reward:** Scaled from baseline ~0.4 to peaks of ~1.0 by step 500.
@@ -104,4 +92,4 @@ The easiest way to re-run the exact GRPO training pipeline is via our Jupyter No
104
 
105
  ### 👥 Team Endurance
106
  * **Shashaank Jain** | GitHub: [@shasshaank](https://github.com/shasshaank) | Email: *[shashaankjain07@gmail.com]*
107
- * **Pranav Pulipati** | GitHub: [@shasshaank](https://github.com/shasshaank) | Email: *[pranavpulipatix@gmail.com]*
 
49
  Our training run clearly demonstrates rapid policy adaptation. The model successfully learned the `OBSERVATION/HYPOTHESIS/ACTION` constraint almost instantly and navigated the tier-2 difficulty bump (step 150) with a textbook drop-and-recover curve.
50
 
51
  ## Training Results
52
+ [W&B Run](https://wandb.ai/shashaankjain07-keshav-memorial-college-of-law/AgentDebuggerEnv/workspace) | [HF Blog](https://huggingface.co/spaces/agentDebugger/AgentDebugger-training-v3/blob/main/Blog.md)
53
 
 
 
 
 
 
 
 
 
 
 
 
 
54
 
55
  * **Format Compliance:** Scaled to 1.0 (max) within 50 steps.
56
  * **Total Reward:** Scaled from baseline ~0.4 to peaks of ~1.0 by step 500.
 
92
 
93
  ### 👥 Team Endurance
94
  * **Shashaank Jain** | GitHub: [@shasshaank](https://github.com/shasshaank) | Email: *[shashaankjain07@gmail.com]*
95
+ * **Pranav Pulipati** | GitHub: [@shasshaank](https://github.com/PulipatiPranav) | Email: *[pranavpulipatix@gmail.com]*