Expose dtype argument (#569)

This commit is contained in:
Lianmin Zheng
2024-06-27 23:30:39 -07:00
committed by GitHub
parent 945aa9beb2
commit badf3fa020
3 changed files with 39 additions and 21 deletions

View File

@@ -120,7 +120,7 @@ class ModelTpServer:
f"[gpu_id={self.gpu_id}] "
f"max_total_num_tokens={self.max_total_num_tokens}, "
f"max_prefill_tokens={self.max_prefill_tokens}, "
f"context_len={self.model_config.context_len}, "
f"context_len={self.model_config.context_len}"
)
if self.tp_rank == 0:
logger.info(