Text Generation
Transformers
English
llama
Inference Endpoints
File size: 2,010 Bytes
b5a3e49
 
589a3c2
 
 
 
b5a3e49
589a3c2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
47587d1
 
 
 
 
 
 
 
 
 
 
 
 
589a3c2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
---
license: bigcode-openrail-m
datasets:
- WizardLM/WizardLM_evol_instruct_70k
language:
- en
---

<font size=5>Here is an example to show how to use model quantized by auto_gptq</font>
```
_4BITS_MODEL_PATH_V1_ = 'GodRain/WizardCoder-15B-V1.1-4bit'

# pip install auto_gptq
from auto_gptq import AutoGPTQForCausalLM
from transformers import AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained(_4BITS_MODEL_PATH_V1_)
model = AutoGPTQForCausalLM.from_quantized(_4BITS_MODEL_PATH_V1_)

out = evaluate("Hello, tell me a story about sun", model=model, tokenizer=tokenizer)
print(out[0].strip())
```

```
def evaluate(
        batch_data,
        tokenizer,
        model,
        temperature=1,
        top_p=0.9,
        top_k=40,
        num_beams=1,
        max_new_tokens=2048,
        **kwargs,
):
    prompts = generate_prompt(batch_data)
    inputs = tokenizer(prompts, return_tensors="pt", max_length=256, truncation=True)
    input_ids = inputs["input_ids"].to(device)
    generation_config = GenerationConfig(
        temperature=temperature,
        top_p=top_p,
        top_k=top_k,
        num_beams=num_beams,
        eos_token_id=tokenizer.eos_token_id,
        pad_token_id=tokenizer.pad_token_id,
        **kwargs,
    )
    with torch.no_grad():
        generation_output = model.generate(
            input_ids=input_ids,
            generation_config=generation_config,
            return_dict_in_generate=True,
            output_scores=True,
            max_new_tokens=max_new_tokens,
        )
    s = generation_output.sequences
    output = tokenizer.batch_decode(s, skip_special_tokens=True)
    return output
```


Citiation:
```
@misc{xu2023wizardlm,
      title={WizardLM: Empowering Large Language Models to Follow Complex Instructions}, 
      author={Can Xu and Qingfeng Sun and Kai Zheng and Xiubo Geng and Pu Zhao and Jiazhan Feng and Chongyang Tao and Daxin Jiang},
      year={2023},
      eprint={2304.12244},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
```