GLM-4.7-Flash-RAZOR-17B-A3B-E32of64
GLM-4.7-Flash with half of its routed experts removed by RAZOR, a training-free expert pruning method. Every MoE layer...
GLM-4.7-Flash with half of its routed experts removed by RAZOR, a training-free expert pruning method. Every MoE layer...
This is the full-precision REAP-288 checkpoint. It is the 288-of-512 expert build spliced directly from the original bf16...
Преобразование модели LiquidAI/LFM2.5-1.2B-Thinking с использованием метода смешанных экспертов (Mixture-of-Experts, MoE), полученное с помощью метода ToMoE (динамическая структурная обрезка...
GGUF builds of Qwen3.6-11B-A3B-Niwaki-4bit-mlx — Qwen3.6-35B-A3B pruned to 19B total / 3.3B active parameters — for llama.cpp and...
A 9–10 GB mixture-of-experts specialist derived from google/gemma-4-26B-A4B-it by pruning 25 % of routed experts with routing-aware REAP...
> A 34%-expert-pruned GLM-5.2 that holds parity with the full unpruned model on a > well-powered real-world eval...
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....
> [!TIP] > Support this work → · X · GitHub · REAP paper · Cerebras REAP 20%...
Эту модель квантовали с использованием квантования смешанной точности oQ. — Тип модели: glm4moelite — Биты: 4 — Размер...
Эта модель была преобразована в формат MLX из cerebras/GLM-4.7-Flash-REAP-23B-A3B с использованием mlx-lm версии 0.30.5. $ uv run mlxlm.convert...