Update app.py
Browse files
app.py
CHANGED
@@ -19,7 +19,7 @@ user_message = "Allie kept track of how many kilometers she walked during the pa
|
|
19 |
model_name = 'TIGER-Lab/StructLM-7B'
|
20 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
21 |
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")
|
22 |
-
model.generation_config = GenerationConfig.from_pretrained(model_name)
|
23 |
# model.generation_config.pad_token_id = model.generation_config.eos_token_id
|
24 |
|
25 |
@spaces.GPU
|
@@ -27,7 +27,6 @@ def predict(user_message, system_message="", assistant_message = "", tabular_dat
|
|
27 |
prompt = f"[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n{assistant_message}\n\n{tabular_data}\n\n\nQuestion:\n\n{user_message}[/INST]"
|
28 |
inputs = tokenizer(prompt, return_tensors='pt', add_special_tokens=True)
|
29 |
input_ids = inputs["input_ids"].to(model.device)
|
30 |
-
|
31 |
output_ids = model.generate(
|
32 |
input_ids,
|
33 |
max_length=input_ids.shape[1] + max_new_tokens,
|
@@ -37,7 +36,6 @@ def predict(user_message, system_message="", assistant_message = "", tabular_dat
|
|
37 |
pad_token_id=tokenizer.eos_token_id,
|
38 |
do_sample=do_sample
|
39 |
)
|
40 |
-
|
41 |
response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
42 |
return response
|
43 |
|
@@ -58,7 +56,7 @@ def main():
|
|
58 |
repetition_penalty = gr.Slider(label="Repetition penalty", value=1.9, minimum=1.0, maximum=2.0)
|
59 |
do_sample = gr.Checkbox(label="Do sample", value=False)
|
60 |
|
61 |
-
output_text = gr.Textbox(label="🐯📏TigerAI-StructLM-7B"
|
62 |
|
63 |
gr.Button("Try🐯📏TigerAI-StructLM").click(
|
64 |
predict,
|
|
|
19 |
model_name = 'TIGER-Lab/StructLM-7B'
|
20 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
21 |
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")
|
22 |
+
# model.generation_config = GenerationConfig.from_pretrained(model_name)
|
23 |
# model.generation_config.pad_token_id = model.generation_config.eos_token_id
|
24 |
|
25 |
@spaces.GPU
|
|
|
27 |
prompt = f"[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n{assistant_message}\n\n{tabular_data}\n\n\nQuestion:\n\n{user_message}[/INST]"
|
28 |
inputs = tokenizer(prompt, return_tensors='pt', add_special_tokens=True)
|
29 |
input_ids = inputs["input_ids"].to(model.device)
|
|
|
30 |
output_ids = model.generate(
|
31 |
input_ids,
|
32 |
max_length=input_ids.shape[1] + max_new_tokens,
|
|
|
36 |
pad_token_id=tokenizer.eos_token_id,
|
37 |
do_sample=do_sample
|
38 |
)
|
|
|
39 |
response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
40 |
return response
|
41 |
|
|
|
56 |
repetition_penalty = gr.Slider(label="Repetition penalty", value=1.9, minimum=1.0, maximum=2.0)
|
57 |
do_sample = gr.Checkbox(label="Do sample", value=False)
|
58 |
|
59 |
+
output_text = gr.Textbox(label="🐯📏TigerAI-StructLM-7B")
|
60 |
|
61 |
gr.Button("Try🐯📏TigerAI-StructLM").click(
|
62 |
predict,
|