deepseek-ai

/

deepseek-coder-33b-instruct

like 459

Follow

DeepSeek 1,318

i think max_position_embeddings should be 4096 here. This value is the original size before scaling.
Reference: https://huggingface.co/docs/transformers/v4.35.2/en/model_doc/llama#transformers.LlamaConfig.rope_scaling