|
Download code/README.md from badr7/rapidchat-data: direct link, hf CLI and curl.
- Browser
- Download file 834 Bytes
-
https://huggingface.co/datasets/badr7/rapidchat-data/resolve/main/code/README.md
- Command line
-
hf download hf://datasets/badr7/rapidchat-data/code/README.md
-
curl -L -o README.md https://huggingface.co/datasets/badr7/rapidchat-data/resolve/main/code/README.md
834 Bytes
rapidchat code
| file | what it does |
|---|---|
vc_bench.py |
builds the re-skinned airline world and the round-1 scenario families |
vc_bench_r2.py, vc_bench_r3.py, vc_bench_r3b.py |
round-2 and round-3 scenario families |
validate_vc_bench.py, validate_r2.py, validate_r3.py |
replay every scenario's gold actions with the real tau2 airline tools and re-derive each answer |
oracle_vc_bench.py, mutation_test.py |
sanity checks: an oracle agent must pass, mutated answers must fail |
vc_install.py |
registers the practice world as a tau2 domain (vc_airline) |
vc_sft.py |
turns passing tau2 conversations into chat-format training examples |
train_lora.py |
LoRA fine-tune (rank 32) with a chunked, checkpointed loss; resumable |
merge.py |
merges an adapter into the base weights |