adam/lulzbot
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

wake up, eric

Browse source
This commit is contained in:
Adam 2022-12-30 22:30:16 -05:00
parent faa542aca4
commit faafd258a3
4 changed files with 64 additions and 26 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

49
src/cartman.py
View file

@ -3,31 +3,34 @@ import json
url = 'https://doordesk.net/chat' url = 'https://doordesk.net/chat'
def cartman_speak(user_message): def cartman_respond(user_message):
message = {'Message': user_message} message = {'Message': user_message}
response = requests.post(url,json.dumps(message)) response = requests.post(url,json.dumps(message))
return response.json().get('Cartman') return response.json().get('Cartman')
from transformers.models.auto.modeling_auto import AutoModelForCausalLM
from transformers.models.auto.tokenization_auto import AutoTokenizer
# from transformers.models.auto.tokenization_auto import AutoTokenizer tokenizer = AutoTokenizer.from_pretrained("microsoft/DialoGPT-large")
# from transformers.models.auto.modeling_auto import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("../chatbots/southpark/cartman/models/output-medium")
# import torch
# def cartman_speak(input_text):
# tokenizer = AutoTokenizer.from_pretrained('microsoft/DialoGPT-large') input_ids = tokenizer(input_text + tokenizer.eos_token, return_tensors="pt").input_ids
# model = AutoModelForCausalLM.from_pretrained('../southpark/output-medium') outputs = model.generate(
# input_ids,
# def cartman_speak(user_message): pad_token_id=tokenizer.eos_token_id,
# new_user_input_ids = tokenizer.encode(user_message + tokenizer.eos_token, return_tensors='pt') max_new_tokens = 200,
# bot_output = new_user_input_ids num_beams = 8,
# bot_input_ids = torch.cat([new_user_input_ids, bot_output]) num_beam_groups = 4,
# bot_output = model.generate( no_repeat_ngram_size=3,
# bot_input_ids, max_length= 200, length_penalty = 1.4,
# pad_token_id=tokenizer.eos_token_id, diversity_penalty = 0,
# no_repeat_ngram_size=3, repetition_penalty = 2.1,
# do_sample=True, early_stopping = True,
# top_k=100,
# top_p=0.7, # do_sample = True,
# temperature=.8 # top_k = 100,
# ) # top_p = 0.7,
# # temperature = 0.8,
# return '{}'.format(tokenizer.decode(bot_output[:,bot_input_ids.shape[-1]:][0], skip_special_tokens=True)) )
return tokenizer.decode(outputs[:, input_ids.shape[-1]:][0], skip_special_tokens=True)

6
test/flantest.py
View file

@ -4,7 +4,7 @@ from transformers.models.t5.modeling_t5 import T5ForConditionalGeneration
device = torch.device("cuda") device = torch.device("cuda")
tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-xl") tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-xl")
model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-xxl").cuda() model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-xl").cuda()
run = True run = True
while run: while run:
@ -12,7 +12,7 @@ while run:
if input_text in 'q': if input_text in 'q':
run = False run = False
break break
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda") input_ids = tokenizer.encode(input_text, return_tensors="pt").to("cuda")
outputs = model.generate(input_ids) outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0])) print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])

18
test/gpt-jt_test.py Normal file
View file

@ -0,0 +1,18 @@
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
device = torch.device("cuda")
tokenizer = AutoTokenizer.from_pretrained("togethercomputer/GPT-JT-6B-v1")
model = AutoModelForCausalLM.from_pretrained("togethercomputer/GPT-JT-6B-v1").cuda()
run = True
while run:
input_text = input('>> ')
if input_text in 'q':
run = False
break
input_ids = tokenizer.encode(input_text, return_tensors="pt").to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])

17
test/gptjtest.py Normal file
View file

@ -0,0 +1,17 @@
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
device = torch.device("cuda")
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-j-6B")
model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-j-6B").cuda()
run = True
while run:
input_text = input('>> ')
if input_text in 'q':
run = False
break
input_ids = tokenizer.encode(input_text, return_tensors="pt").to("cuda")
outputs = model.generate(input_ids)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])