alexkueck commited on
Commit
9f3da8d
·
1 Parent(s): 4e0a704

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +3 -2
app.py CHANGED
@@ -121,7 +121,7 @@ login(token=os.environ["HF_ACCESS_TOKEN"])
121
  base_model = "EleutherAI/gpt-neo-1.3B" #load_8bit = False (in load_tokenizer_and_model)
122
  #base_model = "TheBloke/airoboros-13B-HF" #load_8bit = True
123
  tokenizer,model,device = load_tokenizer_and_model(base_model, False)
124
- tokenizer.add_special_tokens({'pad_token': '[PAD]'})
125
 
126
  dataset_neu = daten_laden("alexkueck/tis")
127
  #dataset_neu = daten_laden("EleutherAI/pile")
@@ -185,7 +185,7 @@ trainer = Trainer(
185
  args=training_args,
186
  train_dataset=lm_datasets["train"],
187
  eval_dataset=lm_datasets["test"],
188
- tokenizer=tokenizer,
189
  #compute_metrics=compute_metrics,
190
  )
191
  print ("trainer")
@@ -227,6 +227,7 @@ print("done")
227
  #Testen des fine-tuned Modells
228
  print("Predict")
229
  predict("Was ist Tis?")
 
230
 
231
  #######################################################################
232
  #Darstellung mit Gradio
 
121
  base_model = "EleutherAI/gpt-neo-1.3B" #load_8bit = False (in load_tokenizer_and_model)
122
  #base_model = "TheBloke/airoboros-13B-HF" #load_8bit = True
123
  tokenizer,model,device = load_tokenizer_and_model(base_model, False)
124
+ #tokenizer.add_special_tokens({'pad_token': '[PAD]'}) #not necessary with fast Toekenizers like GPT2
125
 
126
  dataset_neu = daten_laden("alexkueck/tis")
127
  #dataset_neu = daten_laden("EleutherAI/pile")
 
185
  args=training_args,
186
  train_dataset=lm_datasets["train"],
187
  eval_dataset=lm_datasets["test"],
188
+ #tokenizer=tokenizer,
189
  #compute_metrics=compute_metrics,
190
  )
191
  print ("trainer")
 
227
  #Testen des fine-tuned Modells
228
  print("Predict")
229
  predict("Was ist Tis?")
230
+ print("done")
231
 
232
  #######################################################################
233
  #Darstellung mit Gradio