Ganesh Karbhari
commited on
Commit
•
6c79cf1
1
Parent(s):
b8f046a
Update app.py
Browse files
app.py
CHANGED
@@ -168,54 +168,127 @@
|
|
168 |
|
169 |
|
170 |
|
171 |
-
from huggingface_hub import InferenceClient
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
172 |
import gradio as gr
|
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 |
mychatbot = gr.Chatbot(
|
212 |
avatar_images=["./user.png", "./bot.png"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
|
213 |
|
214 |
-
demo = gr.ChatInterface(fn=
|
215 |
chatbot=mychatbot,
|
216 |
-
title="
|
217 |
retry_btn=None,
|
218 |
undo_btn=None
|
219 |
)
|
220 |
|
221 |
-
demo.queue().launch(show_api=False)
|
|
|
|
168 |
|
169 |
|
170 |
|
171 |
+
# from huggingface_hub import InferenceClient
|
172 |
+
# import gradio as gr
|
173 |
+
# client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.2")
|
174 |
+
|
175 |
+
# def format_prompt(message, history):
|
176 |
+
# prompt = "<s>"
|
177 |
+
# for user_prompt, bot_response in history:
|
178 |
+
# prompt += f"[INST] {user_prompt} [/INST]"
|
179 |
+
# prompt += f" {bot_response}</s> "
|
180 |
+
# prompt += f"[INST] {message} [/INST]"
|
181 |
+
# return prompt
|
182 |
+
|
183 |
+
# def generate(
|
184 |
+
# prompt, history, temperature=0.2, max_new_tokens=3000, top_p=0.95, repetition_penalty=1.0,
|
185 |
+
# ):
|
186 |
+
# temperature = float(temperature)
|
187 |
+
# if temperature < 1e-2:
|
188 |
+
# temperature = 1e-2
|
189 |
+
# top_p = float(top_p)
|
190 |
+
|
191 |
+
# generate_kwargs = dict(
|
192 |
+
# temperature=temperature,
|
193 |
+
# max_new_tokens=max_new_tokens,
|
194 |
+
# top_p=top_p,
|
195 |
+
# repetition_penalty=repetition_penalty,
|
196 |
+
# do_sample=True,
|
197 |
+
# seed=42,
|
198 |
+
# )
|
199 |
+
|
200 |
+
# formatted_prompt = format_prompt(prompt, history)
|
201 |
+
|
202 |
+
# stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
203 |
+
# output = ""
|
204 |
+
|
205 |
+
# for response in stream:
|
206 |
+
# output += response.token.text
|
207 |
+
# yield output
|
208 |
+
# return output
|
209 |
+
|
210 |
+
|
211 |
+
# mychatbot = gr.Chatbot(
|
212 |
+
# avatar_images=["./user.png", "./bot.png"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
|
213 |
+
|
214 |
+
# demo = gr.ChatInterface(fn=generate,
|
215 |
+
# chatbot=mychatbot,
|
216 |
+
# title="Mistral-Chat",
|
217 |
+
# retry_btn=None,
|
218 |
+
# undo_btn=None
|
219 |
+
# )
|
220 |
+
|
221 |
+
# demo.queue().launch(show_api=False)
|
222 |
+
|
223 |
+
|
224 |
+
|
225 |
+
|
226 |
+
|
227 |
+
|
228 |
+
|
229 |
+
|
230 |
+
import boto3
|
231 |
+
import json
|
232 |
+
from botocore.exceptions import ClientError
|
233 |
+
import os
|
234 |
+
|
235 |
+
access_key_id = os.environ['aws_access_key_id']
|
236 |
+
secret_access_key = os.environ['aws_secret_access_key']
|
237 |
import gradio as gr
|
238 |
+
|
239 |
+
bedrock = boto3.client(service_name='bedrock-runtime',region_name='us-east-1',aws_access_key_id=access_key_id,aws_secret_access_key=secret_access_key)
|
240 |
+
|
241 |
+
|
242 |
+
def invoke_llama3_8b(user_message):
|
243 |
+
try:
|
244 |
+
# Set the model ID, e.g., Llama 3 8B Instruct.
|
245 |
+
model_id = "meta.llama3-8b-instruct-v1:0"
|
246 |
+
|
247 |
+
|
248 |
+
# Embed the message in Llama 3's prompt format.
|
249 |
+
prompt = f"""
|
250 |
+
<|begin_of_text|>
|
251 |
+
<|start_header_id|>user<|end_header_id|>
|
252 |
+
{user_message}
|
253 |
+
<|eot_id|>
|
254 |
+
<|start_header_id|>assistant<|end_header_id|>
|
255 |
+
"""
|
256 |
+
|
257 |
+
# Format the request payload using the model's native structure.
|
258 |
+
request = {
|
259 |
+
"prompt": prompt,
|
260 |
+
# Optional inference parameters:
|
261 |
+
"max_gen_len": 1024,
|
262 |
+
"temperature": 0.6,
|
263 |
+
"top_p": 0.9,
|
264 |
+
}
|
265 |
+
|
266 |
+
# Encode and send the request.
|
267 |
+
response = bedrock.invoke_model(body=json.dumps(request), modelId=model_id)
|
268 |
+
|
269 |
+
# Decode the native response body.
|
270 |
+
model_response = json.loads(response["body"].read())
|
271 |
+
|
272 |
+
# Extract and print the generated text.
|
273 |
+
response_text = model_response["generation"]
|
274 |
+
|
275 |
+
return response_text
|
276 |
+
|
277 |
+
except ClientError:
|
278 |
+
print("Couldn't invoke llama3 8B")
|
279 |
+
raise
|
280 |
+
|
281 |
|
282 |
|
283 |
mychatbot = gr.Chatbot(
|
284 |
avatar_images=["./user.png", "./bot.png"], bubble_full_width=False, show_label=False, show_copy_button=True, likeable=True,)
|
285 |
|
286 |
+
demo = gr.ChatInterface(fn=invoke_llama3_8b,
|
287 |
chatbot=mychatbot,
|
288 |
+
title="llama3-Chat",
|
289 |
retry_btn=None,
|
290 |
undo_btn=None
|
291 |
)
|
292 |
|
293 |
+
demo.queue().launch(show_api=False)
|
294 |
+
|