shibing624
commited on
Commit
•
fa56d50
1
Parent(s):
f2a5f61
Update README.md
Browse files
README.md
CHANGED
@@ -114,11 +114,11 @@ from peft import PeftModel
|
|
114 |
from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
|
115 |
|
116 |
|
117 |
-
model = AutoModelForCausalLM.from_pretrained("baichuan-inc/Baichuan-13B-Chat", device_map='auto', trust_remote_code=True)
|
118 |
model.generation_config = GenerationConfig.from_pretrained("baichuan-inc/Baichuan-13B-Chat", trust_remote_code=True)
|
119 |
tokenizer = AutoTokenizer.from_pretrained("baichuan-inc/Baichuan-13B-Chat", trust_remote_code=True)
|
120 |
model = PeftModel.from_pretrained(model, "shibing624/vicuna-baichuan-13b-chat-lora")
|
121 |
-
device =
|
122 |
|
123 |
def generate_prompt(instruction):
|
124 |
return f"""A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.USER: {instruction} ASSISTANT: """
|
@@ -128,11 +128,11 @@ sents = ['一岁宝宝发烧能吃啥药', "who are you?"]
|
|
128 |
for s in sents:
|
129 |
q = generate_prompt(s)
|
130 |
inputs = tokenizer(q, return_tensors="pt")
|
131 |
-
inputs = inputs.to(device
|
132 |
|
133 |
generate_ids = model.generate(
|
134 |
**inputs,
|
135 |
-
max_new_tokens=
|
136 |
)
|
137 |
|
138 |
output = tokenizer.batch_decode(generate_ids, skip_special_tokens=True)[0]
|
@@ -154,6 +154,7 @@ vicuna-baichuan-13b-chat-lora
|
|
154 |
└── adapter_model.bin
|
155 |
```
|
156 |
|
|
|
157 |
|
158 |
### Inference Examples
|
159 |
|
|
|
114 |
from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
|
115 |
|
116 |
|
117 |
+
model = AutoModelForCausalLM.from_pretrained("baichuan-inc/Baichuan-13B-Chat", device_map='auto', torch_dtype=torch.float16, trust_remote_code=True)
|
118 |
model.generation_config = GenerationConfig.from_pretrained("baichuan-inc/Baichuan-13B-Chat", trust_remote_code=True)
|
119 |
tokenizer = AutoTokenizer.from_pretrained("baichuan-inc/Baichuan-13B-Chat", trust_remote_code=True)
|
120 |
model = PeftModel.from_pretrained(model, "shibing624/vicuna-baichuan-13b-chat-lora")
|
121 |
+
device = torch.device(0) if torch.cuda.is_available() else torch.device("cpu")
|
122 |
|
123 |
def generate_prompt(instruction):
|
124 |
return f"""A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.USER: {instruction} ASSISTANT: """
|
|
|
128 |
for s in sents:
|
129 |
q = generate_prompt(s)
|
130 |
inputs = tokenizer(q, return_tensors="pt")
|
131 |
+
inputs = inputs.to(device)
|
132 |
|
133 |
generate_ids = model.generate(
|
134 |
**inputs,
|
135 |
+
max_new_tokens=512,
|
136 |
)
|
137 |
|
138 |
output = tokenizer.batch_decode(generate_ids, skip_special_tokens=True)[0]
|
|
|
154 |
└── adapter_model.bin
|
155 |
```
|
156 |
|
157 |
+
- Inference GPU: 27G
|
158 |
|
159 |
### Inference Examples
|
160 |
|