File size: 7,953 Bytes
b5b75b0 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 |
---
base_model:
- Undi95/Llama-3-Unholy-8B
- Locutusque/llama-3-neural-chat-v1-8b
- ruslanmv/Medical-Llama3-8B-16bit
library_name: transformers
tags:
- mergekit
- merge
- medical
license: other
datasets:
- mlabonne/orpo-dpo-mix-40k
- Open-Orca/SlimOrca-Dedup
- jondurbin/airoboros-3.2
- microsoft/orca-math-word-problems-200k
- m-a-p/Code-Feedback
- MaziyarPanahi/WizardLM_evol_instruct_V2_196k
- ruslanmv/ai-medical-chatbot
model-index:
- name: Medichat-Llama3-8B
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: AI2 Reasoning Challenge (25-Shot)
type: ai2_arc
config: ARC-Challenge
split: test
args:
num_few_shot: 25
metrics:
- type: acc_norm
value: 59.13
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: HellaSwag (10-Shot)
type: hellaswag
split: validation
args:
num_few_shot: 10
metrics:
- type: acc_norm
value: 82.9
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU (5-Shot)
type: cais/mmlu
config: all
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 60.35
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: TruthfulQA (0-shot)
type: truthful_qa
config: multiple_choice
split: validation
args:
num_few_shot: 0
metrics:
- type: mc2
value: 49.65
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: Winogrande (5-shot)
type: winogrande
config: winogrande_xl
split: validation
args:
num_few_shot: 5
metrics:
- type: acc
value: 78.93
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GSM8k (5-shot)
type: gsm8k
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 60.35
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=sethuiyer/Medichat-Llama3-8B
name: Open LLM Leaderboard
language:
- en
---
### Medichat-Llama3-8B
Built upon the powerful LLaMa-3 architecture and fine-tuned on an extensive dataset of health information, this model leverages its vast medical knowledge to offer clear, comprehensive answers.
This model is generally better for accurate and informative responses, particularly for users seeking in-depth medical advice.
The following YAML configuration was used to produce this model:
```yaml
models:
- model: Undi95/Llama-3-Unholy-8B
parameters:
weight: [0.25, 0.35, 0.45, 0.35, 0.25]
density: [0.1, 0.25, 0.5, 0.25, 0.1]
- model: Locutusque/llama-3-neural-chat-v1-8b
- model: ruslanmv/Medical-Llama3-8B-16bit
parameters:
weight: [0.55, 0.45, 0.35, 0.45, 0.55]
density: [0.1, 0.25, 0.5, 0.25, 0.1]
merge_method: dare_ties
base_model: Locutusque/llama-3-neural-chat-v1-8b
parameters:
int8_mask: true
dtype: bfloat16
```
# Comparision Against Dr.Samantha 7B
| Subject | Medichat-Llama3-8B Accuracy (%) | Dr. Samantha Accuracy (%) |
|-------------------------|---------------------------------|---------------------------|
| Clinical Knowledge | 71.70 | 52.83 |
| Medical Genetics | 78.00 | 49.00 |
| Human Aging | 70.40 | 58.29 |
| Human Sexuality | 73.28 | 55.73 |
| College Medicine | 62.43 | 38.73 |
| Anatomy | 64.44 | 41.48 |
| College Biology | 72.22 | 52.08 |
| High School Biology | 77.10 | 53.23 |
| Professional Medicine | 63.97 | 38.73 |
| Nutrition | 73.86 | 50.33 |
| Professional Psychology | 68.95 | 46.57 |
| Virology | 54.22 | 41.57 |
| High School Psychology | 83.67 | 66.60 |
| **Average** | **70.33** | **48.85** |
The current model demonstrates a substantial improvement over the previous [Dr. Samantha](sethuiyer/Dr_Samantha-7b) model in terms of subject-specific knowledge and accuracy.
### Usage:
```python
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
class MedicalAssistant:
def __init__(self, model_name="sethuiyer/Medichat-Llama3-8B", device="cuda"):
self.device = device
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
self.model = AutoModelForCausalLM.from_pretrained(model_name).to(self.device)
self.sys_message = '''
You are an AI Medical Assistant trained on a vast dataset of health information. Please be thorough and
provide an informative answer. If you don't know the answer to a specific medical inquiry, advise seeking professional help.
'''
def format_prompt(self, question):
messages = [
{"role": "system", "content": self.sys_message},
{"role": "user", "content": question}
]
prompt = self.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
return prompt
def generate_response(self, question, max_new_tokens=512):
prompt = self.format_prompt(question)
inputs = self.tokenizer(prompt, return_tensors="pt").to(self.device)
with torch.no_grad():
outputs = self.model.generate(**inputs, max_new_tokens=max_new_tokens, use_cache=True)
answer = self.tokenizer.batch_decode(outputs, skip_special_tokens=True)[0].strip()
return answer
if __name__ == "__main__":
assistant = MedicalAssistant()
question = '''
Symptoms:
Dizziness, headache, and nausea.
What is the differential diagnosis?
'''
response = assistant.generate_response(question)
print(response)
```
## Quants
Thanks to [Quant Factory](https://huggingface.co/QuantFactory), the quantized version of this model is available at [QuantFactory/Medichat-Llama3-8B-GGUF](https://huggingface.co/QuantFactory/Medichat-Llama3-8B-GGUF),
## Ollama
This model is now also available on Ollama. You can use it by running the command ```ollama run monotykamary/medichat-llama3``` in your
terminal. If you have limited computing resources, check out this [video](https://www.youtube.com/watch?v=Qa1h7ygwQq8) to learn how to run it on
a Google Colab backend. |