Views
No views yet
- Layers: 12
- Hidden Size: 768
- Attention Heads: 16 (Query) / 4 (Key-Value)
- FFN Hidden Size: 3,072
- Dropout: 0.1pip install transformers torch1from transformers import AutoTokenizer, AutoModelForCausalLM
2
3# 加载模型（需要设置 trust_remote_code=True）
4model = AutoModelForCausalLM.from_pretrained(
5 "tutututu1998/TinyChat-0.1B",
6 trust_remote_code=True
7)
8
9# 加载 tokenizer
10tokenizer = AutoTokenizer.from_pretrained(
11 "tutututu1998/TinyChat-0.1B",
12 trust_remote_code=True
13)
14
15# 生成文本
16inputs = tokenizer("你好，", return_tensors="pt")
17outputs = model.generate(
18 **inputs,
19 max_length=50,
20 temperature=0.7,
21 do_sample=True
22)
23print(tokenizer.decode(outputs[0], skip_special_tokens=True))1texts = ["你好，", "今天天气", "人工智能"]
2inputs = tokenizer(texts, return_tensors="pt", padding=True)
3outputs = model.generate(**inputs, max_length=50)
4
5for i, output in enumerate(outputs):
6 print(f"输入 {i+1}: {texts[i]}")
7 print(f"输出 {i+1}: {tokenizer.decode(output, skip_special_tokens=True)}")
8 print("-" * 50)1# 首次生成
2inputs = tokenizer("你好，", return_tensors="pt")
3outputs = model.generate(
4 **inputs,
5 max_length=50,
6 use_cache=True # 启用 KV cache
7)1@misc{tinychat2024,
2 author = {Your Name},
3 title = {TinyChat: A Lightweight Language Model},
4 year = {2024},
5 publisher = {HuggingFace},
6 howpublished = {\url{https://huggingface.co/your-username/TinyChat-0.1B}}
7}