File size: 3,028 Bytes
94a4377
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
---
library_name: peft
base_model: Qwen/Qwen2-1.5B-Instruct
pipeline_tag: text-generation
license: apache-2.0
---

# Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->



## Model Details

### Model Description

<!-- Provide a longer summary of what this model is. -->



- **Developed by: hack337**
- **Model type: qwen2**
- **Finetuned from model: Qwen/Qwen2-1.5B-Instruct**

### Model Sources [optional]

<!-- Provide the basic links for the model. -->

- **Repository: https://huggingface.co/Hack337/WavGPT-1.0**
- **Demo: https://huggingface.co/spaces/Hack337/WavGPT**

## How to Get Started with the Model

Use the code below to get started with the model.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
device = "cuda" # the device to load the model onto

model = AutoModelForCausalLM.from_pretrained(
    "Hack337/WavGPT-1.0-merged",
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Hack337/WavGPT-1.0-merged")

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "Вы очень полезный помощник."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(device)

generated_ids = model.generate(
    model_inputs.input_ids,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]

```

Use the code below to get started with the model using NPU.

```python
from transformers import AutoTokenizer, TextStreamer
from intel_npu_acceleration_library import NPUModelForCausalLM
import torch

# Load the NPU-optimized model without LoRA
model = NPUModelForCausalLM.from_pretrained(
    "Hack337/WavGPT-1.0-merged",
    use_cache=True,
    dtype=torch.float16  # Use float16 for the NPU
).eval()

# Load the tokenizer
tokenizer = AutoTokenizer.from_pretrained("Hack337/WavGPT-1.0-merged")
tokenizer.pad_token_id = tokenizer.eos_token_id
streamer = TextStreamer(tokenizer, skip_special_tokens=True)

# Prompt handling
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "Вы очень полезный помощник."},
    {"role": "user", "content": prompt}
]

# Convert to a text format compatible with the model
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
prefix = tokenizer([text], return_tensors="pt")["input_ids"].to("npu")

# Generation configuration
generation_kwargs = dict(
    input_ids=prefix,
    streamer=streamer,
    do_sample=True,
    top_k=50,
    top_p=0.9,
    max_new_tokens=512,
)

# Run inference on the NPU
print("Run inference")
_ = model.generate(**generation_kwargs)

```

- PEFT 0.11.1