All issues

Search results

Enabling LM CACHE failing with glm53-spark-tp2.compose.yaml

Status: Open.

#108 In local-inference-lab/rtx6kpro;

· ian-hailey opened on Sep 29, 2026

glm53-spark-tp2: occasional repetition loops (reasoning and answer) at default sampling

Status: Open.

#107 In local-inference-lab/rtx6kpro;

· mmssix opened on Sep 24, 2026

[Bug] Failed to import Triton kernels (triton_kernels.matmul_ogs) when serving DeepSeek-V4.1-Flash on 4x RTX PRO 6000

Status: Open.

#105 In local-inference-lab/rtx6kpro;

· mathppp opened on Sep 18, 2026

Delayed production CUDA OOM on ds4-vision-jovian-judgement-r9: long-context prefill exceeds profiled peak by ~3 GiB

Status: Open.

#103 In local-inference-lab/rtx6kpro;

· Seth777777777 opened on Sep 15, 2026

Qwen3.8-Flash-Next: could multiple TP1 replicas share the CPU-offloaded PLE table on one node?

Status: Open.

#102 In local-inference-lab/rtx6kpro;

· renehonig opened on Sep 15, 2026

请求增加对nvidia/Qwen3.8-Flash-Next-NVFP4检查点的支持

Status: Open.

#101 In local-inference-lab/rtx6kpro;

· wwwcxc5120 opened on Sep 11, 2026

about models/glm-5.3-flash.md

Status: Open.

#99 In local-inference-lab/rtx6kpro;

· scshsy opened on Sep 7, 2026

[JJ DS4] r9 runtime corrections, qualification and source review checklist

Status: Open.

#95 In local-inference-lab/rtx6kpro;

· voipmonitor opened on Sep 4, 2026

[JJ r1] DeepSeek-V4-Flash source merge contract

Status: Open.

#94 In local-inference-lab/rtx6kpro;

· voipmonitor opened on Sep 3, 2026

Model request: Tencent Hy4-preview (hy_v4) — 770B/49B MoE, gated DSA, 1M ctx

Status: Open.

#90 In local-inference-lab/rtx6kpro;

· hughmadden opened on Aug 30, 2026

[Bug] FlashInfer sparse MLA attention fails with "Unsupported sparse-MLA prefill configuration" during DSpark CUDA Graph capture

Status: Open.

#86 In local-inference-lab/rtx6kpro;

· BrightMoon-000 opened on Aug 28, 2026

请求增加zai-org/GLM-5.3-Flash的支持

Status: Open.

#84 In local-inference-lab/rtx6kpro;

· wwwcxc5120 opened on Aug 27, 2026