making kv_cache_batch_size equivalent to full_batch_size for prefill-only model#687

Merged

quic-hemagnih merged 3 commits intorelease/v1.21.0from

disagg_hotfix_gpt_oss

Dec 23, 2025

Commits on Dec 23, 2025

making kv_cache_batch_size equivalent to full_batch_size for prefill_only model, even though this is not intuitive, being done for vllm interface to be easier
ochougul
committed
moved chunking example script to right dir
ochougul
committed
enhancing readability
ochougul
committed