Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -15,6 +15,16 @@ model_name = "allenai/tulu-2-dpo-13b"
|
|
| 15 |
tokenizer = AutoTokenizer.from_pretrained("allenai/tulu-2-dpo-13b")
|
| 16 |
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
|
| 17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
class TuluChatBot:
|
| 19 |
def __init__(self, model, tokenizer, system_message="You are 🌷Tulu, an AI language model created by Tonic-AI. You are a cautious assistant. You carefully follow instructions. You are helpful and harmless and you follow ethical guidelines and promote positive behavior."):
|
| 20 |
self.model = model
|
|
@@ -30,7 +40,7 @@ class TuluChatBot:
|
|
| 30 |
|
| 31 |
def Tulu(self, user_message, temperature, max_new_tokens, top_p, repetition_penalty, do_sample):
|
| 32 |
prompt = self.format_prompt(user_message)
|
| 33 |
-
inputs = self.tokenizer(prompt, return_tensors='pt', add_special_tokens=
|
| 34 |
input_ids = inputs["input_ids"].to(self.model.device)
|
| 35 |
attention_mask = inputs["attention_mask"].to(self.model.device)
|
| 36 |
|
|
|
|
| 15 |
tokenizer = AutoTokenizer.from_pretrained("allenai/tulu-2-dpo-13b")
|
| 16 |
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
|
| 17 |
|
| 18 |
+
bos_token_id = 1
|
| 19 |
+
eos_token_id = 2
|
| 20 |
+
tokenizer.bos_token_id = bos_token_id
|
| 21 |
+
tokenizer.eos_token_id = eos_token_id
|
| 22 |
+
model.config.bos_token_id = bos_token_id
|
| 23 |
+
model.config.eos_token_id = eos_token_id
|
| 24 |
+
if tokenizer.pad_token is None:
|
| 25 |
+
tokenizer.pad_token = tokenizer.eos_token
|
| 26 |
+
model.config.pad_token_id = tokenizer.convert_tokens_to_ids(tokenizer.pad_token)
|
| 27 |
+
|
| 28 |
class TuluChatBot:
|
| 29 |
def __init__(self, model, tokenizer, system_message="You are 🌷Tulu, an AI language model created by Tonic-AI. You are a cautious assistant. You carefully follow instructions. You are helpful and harmless and you follow ethical guidelines and promote positive behavior."):
|
| 30 |
self.model = model
|
|
|
|
| 40 |
|
| 41 |
def Tulu(self, user_message, temperature, max_new_tokens, top_p, repetition_penalty, do_sample):
|
| 42 |
prompt = self.format_prompt(user_message)
|
| 43 |
+
inputs = self.tokenizer(prompt, return_tensors='pt', add_special_tokens=True)
|
| 44 |
input_ids = inputs["input_ids"].to(self.model.device)
|
| 45 |
attention_mask = inputs["attention_mask"].to(self.model.device)
|
| 46 |
|