File size: 1,698 Bytes
6d90fef
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
library_name: transformers
pipeline_tag: text-generation
inference: true
widget:
- text: Hello!
  example_title: Hello world
  group: Python
---

This model is for debugging. It is randomly initialized using the config from [mistralai/mathstral-7B-v0.1](https://huggingface.co/mistralai/mathstral-7B-v0.1) but with smaller size. 

Codes:
```python
from huggingface_hub import create_repo, upload_folder
from transformers import (
    pipeline,
    set_seed,
    AutoConfig,
    AutoModelForCausalLM,
    AutoTokenizer,
    GenerationConfig,
)
import torch
import transformers
import os

model_id = "mistralai/mathstral-7B-v0.1"
repo_id = "yujiepan/mathstral-v0.1-tiny-random"
save_path = f"/tmp/{repo_id}"

config = AutoConfig.from_pretrained(model_id)
config.hidden_size = 8
config.intermediate_size = 32
config.num_attention_heads = 4
config.num_hidden_layers = 2
config.num_key_value_heads = 2
config.head_dim = 2
print(config)

tokenizer = AutoTokenizer.from_pretrained(model_id)
tokenizer.save_pretrained(save_path)

model = AutoModelForCausalLM.from_config(config, torch_dtype=torch.bfloat16)
model.generation_config = GenerationConfig.from_pretrained(model_id)

set_seed(42)
with torch.no_grad():
    for _, p in sorted(model.named_parameters()):
        torch.nn.init.uniform_(p, -0.1, 0.1)

model.save_pretrained(save_path)

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, do_sample=False, device="cuda")
print(pipe("Hello World!"))

messages = [
    {"role": "system", "content": "You are a robot."},
    {"role": "user", "content": "Hi!"},
]
chatbot = pipeline("text-generation", model=save_path, max_length=1000, max_new_tokens=16)
print(chatbot(messages))
```