Downloads · 30 days
11
5% of all-time downloads
fzkun/minimind3-ascend-moe
minimind3-ascend-moe is a text generation model from fzkun. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
这是一个基于 MiniMind3-Ascend 训练链路导出的 MoE 对话模型,兼容 Transformers 推理方式,适合追求更高容量与更强表达能力的场景。
Downloads · 30 days
11
5% of all-time downloads
All-time downloads
224
Public
Parameters
198M
397 MB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors397 MB · 100%
From the Hugging Face model README
这是一个基于 MiniMind3-Ascend 训练链路导出的 MoE 对话模型,兼容 Transformers 推理方式,适合追求更高容量与更强表达能力的场景。
fzkun/minimind3-ascend-moeQwen3MoeForCausalLMhidden_size = 768num_hidden_layers = 8num_experts = 4num_experts_per_tok = 1仓库中包含:
config.jsongeneration_config.jsonmodel.safetensorstokenizer.jsontokenizer_config.jsonspecial_tokens_map.jsonchat_template.jinjafrom transformers import AutoTokenizer, AutoModelForCausalLM
model_id = "fzkun/minimind3-ascend-moe"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, trust_remote_code=True)
messages = [{"role": "user", "content": "你好"}]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt")
out = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))
评测环境:
npu:0batch_size = 16| ceval | cmmlu | arc | piqa | openbookqa | hellaswag | siqa |
|---|---|---|---|---|---|---|
| 23.77 | 24.88 | 30.30 | 51.63 | 26.00 | 28.58 | 34.08 |
说明:
ceval / cmmlu / arc / piqa / openbookqa / hellaswag 使用 acc_normsocial_iqa 使用 acc