[Feature Request] Qwen 7B model support #684
Description
Qwen 7B is reporting impressive numbers for both English and Chinese.
Qwen is similar to Llama model. Need to add bias params for QKV, and change the tokenizer to tiktoken.
{{ message }}
Qwen 7B is reporting impressive numbers for both English and Chinese.
Qwen is similar to Llama model. Need to add bias params for QKV, and change the tokenizer to tiktoken.