-
Notifications
You must be signed in to change notification settings - Fork 44
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
- Status: Open.#99 In wtdcode/vllm-backport;
- Status: Open.#97 In wtdcode/vllm-backport;
[Bug]: GLM-5.3-Flash degenerates into a constant-token loop with PP=2 + MTP + prefix caching on sm_86 (upstream fix exists: vllm#55506)
bugSomething isn't workingSomething isn't workingStatus: Open.#96 In wtdcode/vllm-backport;[Bug]: CUDA OOM in mhc_post_tilelang (torch.empty_like(residual)) during long-context prefill with DeepSeek-V4.1-Flash, TP=8, DSpark spec decode
bugSomething isn't workingSomething isn't workingStatus: Open.#94 In wtdcode/vllm-backport;- Status: Open.#93 In wtdcode/vllm-backport;
- Status: Open.#91 In wtdcode/vllm-backport;
- Status: Open.#87 In wtdcode/vllm-backport;
- Status: Open.#83 In wtdcode/vllm-backport;
[Bug]: qwen3.8flash bug
bugSomething isn't workingSomething isn't workingStatus: Open.#73 In wtdcode/vllm-backport;- Status: Open.#69 In wtdcode/vllm-backport;
- Status: Open.#68 In wtdcode/vllm-backport;