全站搜索

AI 模型 84759

• certified-scorpion-topo-cbp-qwen-master-GGUF↗

  frankmorales2020

• Qwen3-8B-f16↗

  geoffmunn

• Qwen3.8-27B-NVFP4-RTX5090-LMHead4↗

  gittensor-model-hub

  > ## ⚠️ Superseded > > The main checkpoint now has the NVFP4 `lm_head` **and** the MTP head removed — it is strictly better than this variant (17.92 GB, 2 shards). > > **Use `Qwen3.8-27B-NVFP4-RTX5090` instead.** This repo stays up so existing pins keep working. `Qwen3.8-27B-NVFP4-RTX5090` with **`lm_head` quantized to NVFP4** instead of left in BF16. Every other tensor is bit-identical to the parent. At concurrency 1 this model is **weight-bandwidth bound, not compute bound**. Measured on the parent: 18.80 GiB of resident weights at 81.6 tok/s implies a **1.65 TB/s** read rate against the RTX 5090's 1.79 TB/s spec — about **92% of peak**. Decode is streaming the whole model once per token.

• paligemma-3b-ft-screen2words-448-jax↗

  Google

• opus-mt-en-zls↗

  Helsinki-NLP

• opus-mt-et-en↗

  Helsinki-NLP

• ELYZA-japanese-Llama-2-13b-instruct-neuron↗

  htokoyo

• Cosmos-Reason2-2B-GGUF↗

  hzang

• granite-4.2-8b-bf16-mlx↗

  ibm-granite

• granite-switch-4.1-8b-preview↗

  ibm-granite