Spaces:
Sleeping
Sleeping
BeveledCube
commited on
Update main.py
Browse files
main.py
CHANGED
@@ -5,7 +5,7 @@ from fastapi import FastAPI
|
|
5 |
|
6 |
import os
|
7 |
|
8 |
-
from transformers import GPT2LMHeadModel, GPT2Tokenizer
|
9 |
import torch
|
10 |
|
11 |
app = FastAPI()
|
@@ -15,15 +15,14 @@ name = "microsoft/DialoGPT-small"
|
|
15 |
# microsoft/DialoGPT-medium
|
16 |
# microsoft/DialoGPT-large
|
17 |
|
18 |
-
# PygmalionAI/pygmalion-350m
|
19 |
-
# PygmalionAI/pygmalion-1.3b
|
20 |
-
# PygmalionAI/pygmalion-6b
|
21 |
-
|
22 |
# mistralai/Mixtral-8x7B-Instruct-v0.1
|
23 |
|
24 |
# Load the Hugging Face GPT-2 model and tokenizer
|
25 |
-
model =
|
26 |
-
tokenizer =
|
|
|
|
|
|
|
27 |
|
28 |
class req(BaseModel):
|
29 |
prompt: str
|
@@ -38,16 +37,37 @@ def read_root(data: req):
|
|
38 |
print("Prompt:", data.prompt)
|
39 |
print("Length:", data.length)
|
40 |
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
6 |
import os
|
7 |
|
8 |
+
from transformers import GPT2LMHeadModel, GPT2Tokenizer, AutoModelForCausalLM, AutoTokenizer
|
9 |
import torch
|
10 |
|
11 |
app = FastAPI()
|
|
|
15 |
# microsoft/DialoGPT-medium
|
16 |
# microsoft/DialoGPT-large
|
17 |
|
|
|
|
|
|
|
|
|
18 |
# mistralai/Mixtral-8x7B-Instruct-v0.1
|
19 |
|
20 |
# Load the Hugging Face GPT-2 model and tokenizer
|
21 |
+
model = AutoModelForCausalLM.from_pretrained(name)
|
22 |
+
tokenizer = AutoTokenizer.from_pretrained(name)
|
23 |
+
|
24 |
+
gpt2model = GPT2LMHeadModel.from_pretrained(name)
|
25 |
+
gpt2tokenizer = GPT2Tokenizer.from_pretrained(name)
|
26 |
|
27 |
class req(BaseModel):
|
28 |
prompt: str
|
|
|
37 |
print("Prompt:", data.prompt)
|
38 |
print("Length:", data.length)
|
39 |
|
40 |
+
if name == "microsoft/DialoGPT-small" or name == "microsoft/DialoGPT-medium" or name == "microsoft/DialoGPT-large":
|
41 |
+
# tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-small")
|
42 |
+
# model = AutoModelForCausalLM.from_pretrained("microsoft/DialoGPT-small")
|
43 |
+
|
44 |
+
step = 1
|
45 |
+
|
46 |
+
# encode the new user input, add the eos_token and return a tensor in Pytorch
|
47 |
+
new_user_input_ids = tokenizer.encode(data.prompt + tokenizer.eos_token, return_tensors='pt')
|
48 |
+
|
49 |
+
# append the new user input tokens to the chat history
|
50 |
+
bot_input_ids = torch.cat([chat_history_ids, new_user_input_ids], dim=-1) if step > 0 else new_user_input_ids
|
51 |
+
|
52 |
+
# generated a response while limiting the total chat history to 1000 tokens,
|
53 |
+
chat_history_ids = model.generate(bot_input_ids, max_length=1000, pad_token_id=tokenizer.eos_token_id)
|
54 |
+
|
55 |
+
generated_text = tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)
|
56 |
+
answer_data = { "answer": generated_text }
|
57 |
+
print("Answer:", generated_text)
|
58 |
+
|
59 |
+
return answer_data
|
60 |
+
else:
|
61 |
+
input_text = data.prompt
|
62 |
+
|
63 |
+
# Tokenize the input text
|
64 |
+
input_ids = gpt2tokenizer.encode(input_text, return_tensors="pt")
|
65 |
+
|
66 |
+
# Generate output using the model
|
67 |
+
output_ids = model.generate(input_ids, max_length=data.length, num_beams=5, no_repeat_ngram_size=2)
|
68 |
+
generated_text = gpt2tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
69 |
+
|
70 |
+
answer_data = { "answer": generated_text }
|
71 |
+
print("Answer:", generated_text)
|
72 |
+
|
73 |
+
return answer_data
|