Qwen3.5-76B
> [!TIP] > Support this work → · X · GitHub · REAP paper · Cerebras REAP 40%...
> [!TIP] > Support this work → · X · GitHub · REAP paper · Cerebras REAP 40%...
— Text backbone: NVFP4 — lmhead: higher precision — Vision tower and vision embeddings: higher precision — Audio...
This is the flat-key variant: expert weights are stored under …experts.. (no moe. namespace). Vanilla upstream vllm v0.20.1...
This checkpoint preserves the hybrid Qwen3.5 DeltaNet + softmax architecture and MTP (Multi-Token Prediction) head from the BF16...
TurboQuant TQ3-compressed version of google/gemma-4-26B-A4B-it. Compression: 3-bit TurboQuant with group size 128. Zero calibration data. Weights are decompressed...
GGUF quantizations of 0xSero/gemma-4-19b-a4b-it-REAP, a 30% expert-pruned variant of google/gemma-4-26b-a4b-it using the REAP (Router-weighted Expert Activation Pruning) method....
> [!IMPORTANT] > Naming notice (2026-04-10). The «HLWQ» technique used in this model is being rebranded to HLWQ...
This model is a fine-tuned version of Qwen/Qwen3.5-9B on the godzillaxz/polymarket-trading-data dataset. The following hyperparameters were used during...
PocketPlaning-1 — это специализированная модель планирования агента ИИ, построенная на Qwen3.5-27B. Он точно настроен для интеллектуальной маршрутизации задач,...
使用 Unsloth + TRL 框架,在 100k 中英双语工具调用数据上进行 LoRA SFT,让 Base 模型具备结构化函数调用能力。 所有数据统一为 TRL 兼容的 JSONL 格式,通过 tokenizer.applychattemplate() 自动转为...