Royrotem100 commited on
Commit
16fb26d
·
1 Parent(s): dc3d805

Fix return values in respond function and add photo

Browse files
Files changed (2) hide show
  1. app.py +5 -1
  2. logo111.png +0 -0
app.py CHANGED
@@ -19,6 +19,10 @@ model_name = "dicta-il/dictalm2.0-instruct"
19
  model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16)
20
  tokenizer = AutoTokenizer.from_pretrained(model_name)
21
 
 
 
 
 
22
  @app.route('/chat', methods=['POST'])
23
  def chat():
24
  data = request.json
@@ -33,7 +37,7 @@ def chat():
33
  inputs = tokenizer(user_input, return_tensors='pt', padding=True, truncation=True)
34
  input_ids = inputs['input_ids']
35
  attention_mask = inputs['attention_mask']
36
- outputs = model.generate(input_ids, attention_mask=attention_mask, pad_token_id=tokenizer.eos_token_id)
37
  response_text = tokenizer.decode(outputs[0], skip_special_tokens=True).replace(user_input, '').strip()
38
 
39
  return jsonify({"response": response_text})
 
19
  model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16)
20
  tokenizer = AutoTokenizer.from_pretrained(model_name)
21
 
22
+ # Set the pad_token to eos_token if not already set
23
+ if tokenizer.pad_token is None:
24
+ tokenizer.pad_token = tokenizer.eos_token
25
+
26
  @app.route('/chat', methods=['POST'])
27
  def chat():
28
  data = request.json
 
37
  inputs = tokenizer(user_input, return_tensors='pt', padding=True, truncation=True)
38
  input_ids = inputs['input_ids']
39
  attention_mask = inputs['attention_mask']
40
+ outputs = model.generate(input_ids, attention_mask=attention_mask)
41
  response_text = tokenizer.decode(outputs[0], skip_special_tokens=True).replace(user_input, '').strip()
42
 
43
  return jsonify({"response": response_text})
logo111.png ADDED