20 yrs in jail or $1 million for downloading Chinese models proposed at congress
A bunch of LLMs scheduled to come at end of January were cancelled / delayed
llama.cpp now supports tool calling (OpenAI-compatible)
DeepSeek R1 671B MoE LLM running on Epyc 9374F and 384GB of RAM (llama.cpp + PR #11446, Q4_K_S, real time)
We've been incredibly fortunate with how things have developed over the past year
If Canada cuts off electricity exports to the US in retaliation to tariffs as they’ve suggested is possible, what would happen to VT?
openai can be opening again
Quadro RTX8000 vs RTX4090 vs RTX5090, which is better for Generative Models?
It’s time to lead guys
mistralai/Mistral-Small-24B-Base-2501 · Hugging Face
What's the best 32b ish model we can run locally now? It's hard to keep up.
I want to roast some AI projects today 🤨. If you are not afraid, share your project in the comments!
Deepseek Benchmark 3090 vs 3060
Mistral Q8 GGUF 🍓 fail
What are you *actually* using R1 for?
The Mac M2 Ultra is faster than 2xH100s in running Deepseek R1 IQ1_S.
1.58bit DeepSeek R1 - 131GB Dynamic GGUF
Deepseek 8B models are trash. Was it propaganda?
Nvidia faces $465 billion loss as DeepSeek disrupts AI market, largest in US market history
Don't forget Deepseek 2.5 (and Unsloth)
I want to search through all my books (fewer than 100). What is the best tool with the largest context window for this?
Who would you prefer to get to AGI first?
DeepSeek is running inference on the new home Chinese chips made by Huawei, the 910C
1 Million Token Context Length 🔥
What setup would you need to run Deepseek r1 locally?
