| 関連文書 |
| 1 |
Zero recycled KV blocks for FullAttention models (#39146) AjAnubolu/vllm@1ad6786 GitHub |
https://github.com/AjAnubolu/vllm/commit/1ad67864c0c20f167929e64c875f5c28e1aad9fd
|
| 2 |
[Bugfix] Zero recycled KV cache blocks for FullAttention models by AjAnubolu Pull Request #39283 vllm-project/vllm GitHub |
https://github.com/vllm-project/vllm/pull/39283
|
| 3 |
[Bug]: KV block corruption in base scheduler, Non-deterministic output at temperature=0 without prefix caching Issue #39146 vllm-project/vllm (https://github.com/vllm-project/vllm/issues/39146#issue-4215090365) |
https://github.com/vllm-project/vllm/issues/39146#issue-4215090365
|
| 4 |
[Bug]: KV block corruption in base scheduler, Non-deterministic output at temperature=0 without prefix caching Issue #39146 vllm-project/vllm (https://github.com/vllm-project/vllm/issues/39146) |
https://github.com/vllm-project/vllm/issues/39146
|