Post
Good news, llama.cpp seems to be close to supporting MTP on qwen models. Bad news, every single gguf will have to be redone when it is.
Matthew Andrews
BlueNipples
AI & ML interests
None yet
Recent Activity
new activity 2 minutes ago
mistralai/Mistral-Medium-3.5-128B:Finally dense 100b+! liked a model 6 days ago
mistralai/Mistral-Medium-3.5-128BOrganizations
posted an update 9 minutes ago
posted an update about 1 year ago
Post
664
I'll never understand why people are merging reasoning models with non-reasoning models. It's worse, every time.
You got to train reasoning on reasoning data, and merge reasoning, with reasoning.
You got to train reasoning on reasoning data, and merge reasoning, with reasoning.
posted an update over 1 year ago
Post
605
arcee-ai/Virtuoso-Lite is really good. That's all lol.
If this is prompting with special code, forgive my dummy question, how does one turn that into a usable fine tune? Using the layers to mass generate DPO pairs for a detox dataset?