Skip to content
#

cpu-offloading

Star

Here are 24 public repositories matching this topic...

Language: All
Filter by language
All 24 Python 22 Markdown 1
Sort: Most stars
Sort options
Most stars Fewest stars Most forks Fewest forks Recently updated Least recently updated

DominikBucko / qwen38-flash-next-2x3090

Star 118

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 15, 2026
  • Python

DominikBucko / qwen38-flash-next-3090

Star 8

Run Qwen3.8-Flash-Next on ONE RTX 3090 (24 GB) + 64 GB RAM: 128K context, up to 2,100 tok/s prefill, 43–51 tok/s decode. vLLM runtime with hot MoE experts on the GPU and cold experts computed on the CPU, INT8 KV cache, Docker, OpenAI-compatible API.

moe mixture-of-experts single-gpu rtx3090 int4 llm long-context vllm local-llm llm-inference qwen kv-cache-quantization openai-compatible-api consumer-gpu rtx-3090 cpu-offloading qwen38 qwen3-8-flash-next 128k-context 24gb-vram
  • Updated Sep 30, 2026
  • Python

RobertKaraca9652 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

ngarcia14 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 14, 2026
  • Python

cooper65 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

somjai48 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

jcelik672 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 15, 2026
  • Python

NikolaiHashimoto / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 16, 2026
  • Python

Janetg34100 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 16, 2026
  • Python

manfred01709 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 15, 2026
  • Python

agnieszkayamamoto11513 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

ghasegawa4059 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

MartinaKowalski2970 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

AchilleasMemon / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026
  • Python

Diaz9949 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 14, 2026
  • Python

virginiah813 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 17, 2026

Luuk038 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 14, 2026
  • Python

rRichardson169 / qwen38-flash-next-2x3090

Star 0

Qwen3.8-Flash-Next on 2× RTX 3090: with 128 GB RAM up to 4,191 tok/s prefill · 111.5 tok/s decode (131K prompt), full 256K window at 2,865 tok/s; with 64 GB RAM 3,410 tok/s prefill · 84 tok/s decode. New: opt-in uncensored mode (runtime abliteration, no new weights).

moe quantization mtp multi-gpu dual-gpu rtx3090 int4 fp8 vllm local-llm llm-inference qwen speculative-decoding rtx4090 rtx-3090 cpu-offloading qwen3-8 qwen38 qwen3-8-flash-next 256k-context
  • Updated Oct 16, 2026
  • Python

Add this topic to your repo

To associate your repository with the cpu-offloading topic, visit your repo's landing page and select "manage topics."

Learn more