GLM 5.2 Nano ( 4 to 20 B params) would be good for edge hardware
please.. should fit on 16GB Vram
I agree, minimax or glm should enter small llm race, qwen recently left, so there is small competition there, mistral stopped working on small models for now, gemma is only competitor in the future I would worry about and others are too far behind, we need new small models
I agree, minimax or glm should enter small llm race, qwen recently left, so there is small competition there, mistral stopped working on small models for now, gemma is only competitor in the future I would worry about and others are too far behind, we need new small models
Why qwen left? Is there any chance to get qwen 3.6 4-9b?
I agree, minimax or glm should enter small llm race, qwen recently left, so there is small competition there, mistral stopped working on small models for now, gemma is only competitor in the future I would worry about and others are too far behind, we need new small models
What about MiniCPM?
I agree, minimax or glm should enter small llm race, qwen recently left, so there is small competition there, mistral stopped working on small models for now, gemma is only competitor in the future I would worry about and others are too far behind, we need new small models
Why qwen left? Is there any chance to get qwen 3.6 4-9b?
Fine tune it yourself on public cpt data