5 Commits

Author	SHA1	Message	Date
badstorm	56bb7b2e43	Force PyTorch's default NCCL process-group timeout in vLLM's ROCm image	2026-08-02 23:03:56 +02:00
badstorm	624e79aaa6	Standardize llama.cpp env vars and binary handling, add vLLM env passthrough	2026-08-02 18:53:14 +02:00
badstorm	cded740792	Add single-node mode, tool/reasoning calling, and TheRock-based RCCL build for vLLM	2026-08-02 12:37:30 +02:00
badstorm	71a3a22d2e	Fix builder image	2026-08-01 23:37:43 +02:00
badstorm	d21a7780a2	Add vllm container	2026-08-01 23:33:51 +02:00