Search results

为什么一直在用虚拟显存

Status: Open.

#667 In ztxz16/fastllm;

· EthanPeng-2048 opened on May 4, 2026

【Bug】DeepSeek-V4-Flash-FP8-Q4 使用 multicuda:0,1 输出乱码，单卡正常

Status: Open.

#666 In ztxz16/fastllm;

· haixinliu7864-alt opened on May 3, 2026

need support deepseek v4

Status: Open.

#665 In ztxz16/fastllm;

· WangHHY19931001 opened on Apr 27, 2026

Why I can't build ROCm code in my machine?

Status: Open.

#664 In ztxz16/fastllm;

· MaoxinYee opened on Apr 23, 2026

Please update the PIP version

Status: Open.

#658 In ztxz16/fastllm;

· artisankat2000 opened on Apr 2, 2026

Qwen3.5-397B FP8 HF load can consume nearly 2x host RAM and OOM before startup

Status: Open.

#657 In ztxz16/fastllm;

· papersail opened on Mar 29, 2026

The nightly build throws a metadata error!?

Status: Open.

#655 In ztxz16/fastllm;

· wingraver opened on Mar 21, 2026

fastllm是否可以Qwen3-Embedding-4B这个模型的推理？

Status: Open.

#654 In ztxz16/fastllm;

· jimmy-kuo opened on Mar 12, 2026

Issue with Model Qwen3.5 Dense FP8

Status: Open.

#653 In ztxz16/fastllm;

· lovedheart opened on Mar 10, 2026

Unsupported model type qwen3_5_moe

Status: Open.

#652 In ztxz16/fastllm;

· lovedheart opened on Mar 10, 2026

似乎提供的openai兼容接口有缺陷？

Status: Open.

#648 In ztxz16/fastllm;

· pdwfree opened on Mar 4, 2026

Dockerfilec中无setup.py错误问题

Status: Open.

#645 In ztxz16/fastllm;

· skylerma1106 opened on Feb 11, 2026