Views
No views yet
| V3-BF16 | V4-BF16 | |
|---|---|---|
| Context Length | 2048 | 4096 (Handles longer ALFWorld trajectories) |
| Dataset | Standard SFT | Iron Guard (171 strictly curated high-quality trajectories) |
| Training Method | Validation enabled | Optimized SFT (Batch 1, GradAcc 4, No validation overhead) |
| Target Focus | General Agent | Aggregations & Complex Planning (SQL SUM/COUNT, long navigation) |
0.192 through 1500 intensive iterations on Mac M4.max_seq_length to 4096.batch_size=1, grad_accumulation=4, and grad_checkpointing.1# Evaluate directly in BF16 format (if within timeout limits)
2cd ~/AgentBench
3./update_model.sh satoyutaka/Qwen2.5-7B-AgentBench-V4-BF16
4python3 -m src.assigner -c configs/assignments/dbbench-vllm.yaml
5python3 -m src.assigner -c configs/assignments/alfworld-vllm.yamlmax_seq_lengthを4096に倍増しました。batch_size=1, grad_accumulation=4, grad_checkpointing を駆使し、MacのOOMエラーを克服しました。0.192)。