can i fit moe qwen3.5 in 10gb vram? since thats already risky, lets yolo and use claude distil too. 0.65 compression this time. my original goal was 8gb vram but i mathed wrong somewhere. that fits fine on my gpro x080 but not single gpu in the radeon v340l. maybe ill give it another attempt if it turns out not terrible. test results: — snake clone in single html file: fail even after 8 iterations. — car wash test: pushing the car is «Possible but exhausting» > 📢 Release Note > Build Environment Upgrades: > — Fine-tuning Framework: Unsloth 2026.3.3 > — Core Dependencies: Transformers 5.2.0 > — Compared to the original model, autonomy and stability are significantly improved. Qwen3.5-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled is a highly capable reasoning model fine-tuned on top of the powerful Qwen3.5 architecture. The model’s core directive is to leverage state-of-the-art Chain-of-Thought (CoT) distillation primarily sourced from Claude-4.6 Opus interactions. Through Supervised Fine-Tuning (SFT) focusing specifically on structured reasoning logic, this model excels in breaking down complex user problems, planning step-by-step methodologies within strictly formatted tags, and…
Модальности:
Генерация текста
Области применения:
Логика и рассуждение Диалог / чат
Задача: Генерация текста
Автор: tvall43
Теги: qwen3_5_moe_text, image-text-to-text, text-generation-inference, unsloth, qwen3_5_moe, qwen, qwen3.5, reasoning
Лайков: 4 | Загрузок: 68
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
