File size: 4,727 Bytes
958a6aa
 
63880d2
 
 
 
 
 
 
958a6aa
63880d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
958a6aa
a937824
958a6aa
 
 
 
 
 
63880d2
 
 
 
958a6aa
 
 
 
 
63880d2
958a6aa
 
63880d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
958a6aa
63880d2
 
 
 
 
 
 
 
 
 
 
 
958a6aa
63880d2
 
 
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
119
120
121
122
123
124
125
126
127
128
129
---
library_name: transformers
tags:
- medical
license: cc-by-nc-sa-4.0
language:
- en
pipeline_tag: text-generation
base_model: Intel/neural-chat-7b-v3-1
---
## InMD-X: Large Language Models for Internal Medicine Doctors
We introduce InMD-X, a collection of
multiple large language models specifically designed
to cater to the unique characteristics and demands
of Internal Medicine Doctors (IMD). InMD-X represents
a groundbreaking development in natural language
processing, offering a suite of language models
fine-tuned for various aspects of the internal medicine
field. These models encompass a wide range of medical
sub-specialties, enabling IMDs to perform more
efficient and accurate research, diagnosis, and documentation.
InMD-X’s versatility and adaptability
make it a valuable tool for improving the healthcare
industry, enhancing communication between healthcare
professionals, and advancing medical research.
Each model within InMD-X is meticulously tailored
to address specific challenges faced by IMDs, ensuring
the highest level of precision and comprehensiveness
in clinical text analysis and decision support. 

(This model card is for the INFECTIOUS DISEASES subspecialty.)


### Model Description

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

- **Model type:** [CausalLM]
- **Language(s) (NLP):** [English]
- **License:** [CC-BY-NC-SA](https://creativecommons.org/licenses/by-nc-sa/4.0/)
- **Finetuned from model [optional]:** [Intel/neural-chat-7b-v3-1](https://huggingface.co/Intel/neural-chat-7b-v3-1)

### Model Sources [optional]

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

- **Paper [optional]:** [InMD-X](http://arxiv.org/abs/2402.11883)

## Uses
```python
import torch
from peft import PeftModel, PeftConfig
from transformers import AutoModelForCausalLM, AutoTokenizer
peft_model_id = "InMedData/InMD-X-INF"
config = PeftConfig.from_pretrained(peft_model_id)
model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=True, load_in_8bit=True, device_map='auto')
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
# Load the Lora model
model = PeftModel.from_pretrained(model, peft_model_id)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    tokenizer = tokenizer,
    device_map="auto" # if you have GPU
)
def inference(pipeline, Qustion,answer_only = False):
    sequences = pipeline("Answer the next question in one sentence.\n"+
    Qustion,
    do_sample=True, 
    top_k=10,
    top_p = 0.9,
    temperature = 0.2,
    num_return_sequences=1,
    eos_token_id=tokenizer.eos_token_id,
    max_length=500, # can increase the length of sequence
    )
    
    Answers = []
    for seq in sequences:
        
        Answer = seq['generated_text'].split(Qustion)[-1].replace("\n","")
        Answers.append(Answer)
    return Answers
        
question = 'What is the association between long-term beta-blocker use after myocardial infarction (MI) and the risk of reinfarction and death?'
answers = inference(pipeline, question)
print(answers)
```


### List of LoRA config 
based on [Parameter-Efficient Fine-Tuning (PEFT)](https://github.com/huggingface/peft)

Parameter | PT | SFT 
:------:| :------:| :------:
r | 8 | 8 
lora alpha | 32 | 32 
lora dropout | 0.05 | 0.05  
target | q, k, v, o,up, down, gate | q, k, v, o,up,down, gate 


### List of Training arguments
based on [Transformer Reinforcement Learning (TRL)](https://github.com/huggingface/trl)

Parameter | PT | SFT 
:------:| :------:| :------:
train epochs | 3 | 1 
per device train batch size | 1 | 1 
optimizer | adamw_hf | adamw_hf
evaluation strategy | no | no
learning_rate | 1e-4 | 1e-4
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Experimental setup
- **Ubuntu 22.04.3 LTS**
- **GPU - NVIDIA A100(40GB)**
- **Python**: 3.10.12
- **Pytorch**:2.1.1+cu118
- **Transformer**:4.37.0.dev0
## Limitations
InMD-X consists of a collection of segmented models. The integration of the models has not yet been fully accomplished, resulting in each model being fragmented.
Due to the absence of benchmarks, the segmented models have not been adequately evaluated. Future research will involve the development of new benchmarks and the integration of models to facilitate an objective evaluation.
## Non-commercial use
These models are available exclusively for research purposes and are not intended for commercial use.
<!-- ## Citation
**BibTeX:**
-->
## INMED DATA
INMED DATA is developing large language models (LLMs) specifically tailored for medical applications. For more information, please visit our website [TBD].