-
Notifications
You must be signed in to change notification settings - Fork 11
/
Copy pathchat.py
123 lines (112 loc) · 5.09 KB
/
chat.py
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
"""
Chat with a trained model
"""
import os
import pickle
from contextlib import nullcontext
import torch
import tiktoken
from model import GPTConfig, GPT
import requests
# -----------------------------------------------------------------------------
init_from = 'resume'
out_dir = 'out-chat' # where finetuned model lives
num_samples = 1 # no samples. 1 for 1 chat at a time
max_new_tokens = 100
temperature = 0.8
top_k = 5 # retain only the top_k most likely tokens, clamp others to have 0 probability
device = 'cuda' # examples: 'cpu', 'cuda', 'cuda:0', 'cuda:1', etc.
dtype = 'bfloat16' if torch.cuda.is_available() and torch.cuda.is_bf16_supported() else 'float16' # 'float32' or 'bfloat16' or 'float16'
compile = True # use PyTorch 2.0 to compile the model to be faster
context="<human>Hello, how are you?<endOfText><bot>Thanks, Im good, what about you?<endOfText><human>Im great thanks, My names James, and I'm from the UK, wbu?<endOfText><bot>Hi James, I'm Conner, and im from america. <endOftext>" # a little context for better chat responses
exec(open('configurator.py').read()) # overrides from command line, only for out_dir location, if you store the ckpt.pt elsewhere, like gdrive, to escape finetuning everytime you run the colab
# -----------------------------------------------------------------------------
torch.backends.cuda.matmul.allow_tf32 = True # allow tf32 on matmul
torch.backends.cudnn.allow_tf32 = True # allow tf32 on cudnn
device_type = 'cuda' if 'cuda' in device else 'cpu' # for later use in torch.autocast
ptdtype = {'float32': torch.float32, 'bfloat16': torch.bfloat16, 'float16': torch.float16}[dtype]
ctx = nullcontext() if device_type == 'cpu' else torch.amp.autocast(device_type=device_type, dtype=ptdtype)
def download_ckpt(url):
response = requests.get(url)
if response.status_code == 200:
with open('ckpt.pt', 'wb') as f:
f.write(response.content)
else:
print('Error downloading file:', response.status_code)
# gets model
# init from a model saved in a specific directory
if init_from == 'huggingface':
if os.path.isfile('ckpt.pt'):
# init from huggingface model
ckpt_path = 'ckpt.pt'
checkpoint = torch.load(ckpt_path, map_location=device)
gptconf = GPTConfig(**checkpoint['model_args'])
model = GPT(gptconf)
state_dict = checkpoint['model']
unwanted_prefix = '_orig_mod.'
for k,v in list(state_dict.items()):
if k.startswith(unwanted_prefix):
state_dict[k[len(unwanted_prefix):]] = state_dict.pop(k)
model.load_state_dict(state_dict)
else:
# init from huggingface model
download_ckpt('https://huggingface.co/VatsaDev/ChatGpt-nano/resolve/main/ckpt.pt')
ckpt_path = 'ckpt.pt'
checkpoint = torch.load(ckpt_path, map_location=device)
gptconf = GPTConfig(**checkpoint['model_args'])
model = GPT(gptconf)
state_dict = checkpoint['model']
unwanted_prefix = '_orig_mod.'
for k,v in list(state_dict.items()):
if k.startswith(unwanted_prefix):
state_dict[k[len(unwanted_prefix):]] = state_dict.pop(k)
model.load_state_dict(state_dict)
elif init_from == 'resume':
ckpt_path = os.path.join(out_dir, 'ckpt.pt')
checkpoint = torch.load(ckpt_path, map_location=device)
gptconf = GPTConfig(**checkpoint['model_args'])
model = GPT(gptconf)
state_dict = checkpoint['model']
unwanted_prefix = '_orig_mod.'
for k,v in list(state_dict.items()):
if k.startswith(unwanted_prefix):
state_dict[k[len(unwanted_prefix):]] = state_dict.pop(k)
model.load_state_dict(state_dict)
model.eval()
model.to(device)
if compile:
model = torch.compile(model) # requires PyTorch 2.0 (optional)
# gpt-2 encodings
print("loading GPT-2 encodings...")
enc = tiktoken.get_encoding("gpt2")
encode = lambda s: enc.encode(s, allowed_special={"<|endoftext|>"})
decode = lambda l: enc.decode(l)
def respond(input, samples): # generation function
x = (torch.tensor(encode(input), dtype=torch.long, device=device)[None, ...])
with torch.no_grad():
with ctx:
for k in range(samples):
generated = model.generate(x, max_new_tokens, temperature=temperature, top_k=top_k)
output = decode(generated[0].tolist())
# replace context
output = output.replace(input,'')
# remove any human response
output = output.partition('<human>')
# if the bot has anything left afterwards, the endOfText token is put to use
output_text = output[0].rpartition('<endOftext>')
output_text = output[0] + output[1]
# label removing
output_text = output_text.replace('<human>',' ')
output_text = output_text.replace('<bot>',' ')
output_text = output_text.replace('<endOfText>',' ')
return output_text
# chat loop
while True:
# get input from user
start_input = input('User: ')
start = '<human>'+start_input+'<endOfText><bot>'
# context
context=context+start
out = respond(context, num_samples)
context=context+out+'<endOfText>'
print('Bot: '+ out)