Jagat Kiran

jagat-primitive-org's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

New activity in primitive-ai/Qwen3.8-Flash-Next-PLE-quant about 1 month ago

Optional resident mode for the quantized tables (VLLM_PLE_QUANT_RESIDENT)

5

#2 opened about 1 month ago by

mgabor3141

New activity in primitive-ai/Qwen3.8-Flash-Next-NVFP4 about 1 month ago

Field report: PLE-quant assets + SSD-Stream shell on a 62GB-RAM RTX 6000 Pro, KV pool up to 1,111,168

4

#2 opened about 1 month ago by

AntigravityAI

New activity in primitive-ai/Qwen3.8-Flash-Next-mixed-NVFP4-FP8 about 1 month ago

Benchmarked this checkpoint + PLE INT4 overlay on 1× RTX PRO 6000 Blackwell (96 GB) + 64 GB host RAM — vs ExLlamaV3 and vs Qwen3.8-27B

6

#3 opened about 1 month ago by

Marzero

Error when using --moe-backend marlin per the 09-03 suggestion

1

#2 opened about 1 month ago by

timbuksu

New activity in primitive-ai/Qwen3.8-Flash-Next-mixed-NVFP4-FP8 about 2 months ago

vLLM hangs post-CUDA graph capture with repeated shm_broadcast timeout errors after patches applied

4

#1 opened about 2 months ago by

sebastiancretu

New activity in primitive-ai/Qwen3.8-Flash-Next-NVFP4 about 2 months ago

what about ram

2

#1 opened about 2 months ago by

xuguowong

New activity in primitive-ai/Qwen3.8-27B-mixed-NVFP4-FP8 about 2 months ago

DFlash2?

1

#1 opened about 2 months ago by

andrew-stanton

New activity in primitive-ai/Ornith-1.5-35B-A3B-mixed-NVFP4-FP8 about 2 months ago

MergedColumnParallelLinear has no attribute 'data'

1

#1 opened about 2 months ago by

flaviusburca