Update app.py

app.py

@@ -17,21 +17,13 @@ import torch
 from transformers import GPT2Tokenizer, GPT2LMHeadModel
 from torch.nn.functional import softmax
 import numpy as np
-
+from huggingface_hub import login
 
 
 #url = "https://github.com/simonepri/lm-scorer/tree/master/lm_scorer/models"
 #resp = requests.get(url)
 
 from sentence_transformers import SentenceTransformer, util
-#from sentence_transformers import SentenceTransformer, util
-#from sklearn.metrics.pairwise import cosine_similarity
-#from lm_scorer.models.auto import AutoLMScorer as LMScorer
-#from sentence_transformers import SentenceTransformer, util
-#from sklearn.metrics.pairwise import cosine_similarity
-
-#device = "cuda:0" if torch.cuda.is_available() else "cpu"
-#model_sts = gr.Interface.load('huggingface/sentence-transformers/stsb-distilbert-base') 
 
 #model_sts = SentenceTransformer('stsb-distilbert-base')
 model_sts = SentenceTransformer('roberta-large-nli-stsb-mean-tokens')
@@ -43,79 +35,33 @@ from transformers import GPT2Tokenizer, GPT2LMHeadModel
 import numpy as np
 import re
 
-# def Sort_Tuple(tup):  
-  
-# 	# (Sorts in descending order)  
-# 	tup.sort(key = lambda x: x[1])  
-# 	return tup[::-1]
-
 
-# def softmax(x):
-# 	exps = np.exp(x)
-# 	return np.divide(exps, np.sum(exps))
-	
 
 def get_sim(x):
     x =  str(x)[1:-1]
     x =  str(x)[1:-1]
     return x
      
-# Load pre-trained model 
-
-# model = GPT2LMHeadModel.from_pretrained('gpt2', output_hidden_states = True, output_attentions = True)
-
-# #model  =  gr.Interface.load('huggingface/distilgpt2', output_hidden_states = True, output_attentions = True)
-
-# #model.eval()
-# #tokenizer =  gr.Interface.load('huggingface/distilgpt2')
-
-# tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
-# #tokenizer = GPT2Tokenizer.from_pretrained('distilgpt2')
-
-
-# def cloze_prob(text):
-
-# 	whole_text_encoding = tokenizer.encode(text)
-# 	# Parse out the stem of the whole sentence (i.e., the part leading up to but not including the critical word)
-# 	text_list = text.split()
-# 	stem = ' '.join(text_list[:-1])
-# 	stem_encoding = tokenizer.encode(stem)
-# 	# cw_encoding is just the difference between whole_text_encoding and stem_encoding
-# 	# note: this might not correspond exactly to the word itself
-# 	cw_encoding = whole_text_encoding[len(stem_encoding):]
-# 	# Run the entire sentence through the model. Then go "back in time" to look at what the model predicted for each token, starting at the stem.
-# 	# Put the whole text encoding into a tensor, and get the model's comprehensive output
-# 	tokens_tensor = torch.tensor([whole_text_encoding])
-	
-# 	with torch.no_grad():
-# 		outputs = model(tokens_tensor)
-# 		predictions = outputs[0]   
-
-# 	logprobs = []
-# 	# start at the stem and get downstream probabilities incrementally from the model(see above)
-# 	start = -1-len(cw_encoding)
-# 	for j in range(start,-1,1):
-# 			raw_output = []
-# 			for i in predictions[-1][j]:
-# 					raw_output.append(i.item())
-	
-# 			logprobs.append(np.log(softmax(raw_output)))
-			
-# 	# if the critical word is three tokens long, the raw_probabilities should look something like this:
-# 	# [ [0.412, 0.001, ... ] ,[0.213, 0.004, ...], [0.002,0.001, 0.93 ...]]
-# 	# Then for the i'th token we want to find its associated probability
-# 	# this is just: raw_probabilities[i][token_index]
-# 	conditional_probs = []
-# 	for cw,prob in zip(cw_encoding,logprobs):
-# 			conditional_probs.append(prob[cw])
-# 	# now that you have all the relevant probabilities, return their product.
-# 	# This is the probability of the critical word given the context before it.
-
-# 	return np.exp(np.sum(conditional_probs))
-
-
-tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
-model = GPT2LMHeadModel.from_pretrained('gpt2')
+
+
+print(os.getenv('HF_token'))
+hf_api_token = os.getenv("HF_token")  # For sensitive secrets
+#app_mode = os.getenv("APP_MODE")  # For public variables
+
+
+access_token = hf_api_token
+#print(login(token = access_token))
+
+
+tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.2-1B")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-1B")
+
+
+
+
+
+#tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
+#model = GPT2LMHeadModel.from_pretrained('gpt2')
 
 def sentence_prob_mean(text):
     # Tokenize the input text and add special tokens