New model to play with:
Deepseek-coder (llama based) is trained from scratch on 87% code and 13% natural language (English / Chinese)
Pre-trained using 2T tokens
Try it 👇
1.3B:
ollama run deepseek-coder
6.7B:
ollama run deepseek-coder:6.7b
33B:
ollama run deepseek-coder:33b
https://t.co/6hOG2ScM6N
No innings comes close to this, for me. Unreal, top batting, hits all around the park, what intent and in such a vital game. Thank you for this @ImRaina! 💛