Files
StackFlow/projects/llm_framework/main_llm/models/mode_qwen2.5-0.5B-prefill-20e.json
2025-04-16 18:34:33 +08:00

36 lines
1.2 KiB
JSON

{
"mode":"qwen2.5-0.5B-prefill-20e",
"type":"llm",
"homepage":"https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct",
"compile_flage":"pulsar2 llm_build --input_path Qwen/Qwen2-0.5B-Instruct/ --output_path Qwen/Qwen2-0.5B-w8a16/ --kv_cache_len 1023 --hidden_state_type bf16 --prefill_len 128 --chip AX620E;./tools/embed_process.sh Qwen/Qwen2-0.5B-Instruct/ Qwen/Qwen2-0.5B-w8a16/",
"pulsar_version":"3.4-983bb35e",
"capabilities":[
"text_generation",
"chat"
],
"input_type":[
"llm.utf-8",
"llm.utf-8.stream",
"llm.chat_completion",
"llm.chat_completion.stream"
],
"output_type":[
"llm.utf-8",
"llm.utf-8.stream"
],
"mode_param":{
"tokenizer_type":1,
"filename_tokenizer_model":"qwen.tiktoken",
"filename_tokens_embed":"model.embed_tokens.weight.bfloat16.bin",
"filename_post_axmodel":"qwen2_post.axmodel",
"template_filename_axmodel":"qwen2_p128_l%d_together.axmodel",
"b_use_topk":false,
"b_bos":false,
"b_eos":false,
"axmodel_num":24,
"tokens_embed_num":151936,
"tokens_embed_size":896,
"b_use_mmap_load_embed":true,
"b_dynamic_load_axmodel_layer":false
}
}