Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,6 +1,7 @@
|
|
| 1 |
import torch
|
| 2 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 3 |
import time
|
|
|
|
| 4 |
|
| 5 |
def generate_prompt(instruction, input=""):
|
| 6 |
instruction = instruction.strip().replace('\r\n','\n').replace('\n\n','\n')
|
|
@@ -25,10 +26,10 @@ model_path = "models/rwkv-6-world-1b6/" # Path to your local model directory
|
|
| 25 |
model = AutoModelForCausalLM.from_pretrained(
|
| 26 |
model_path,
|
| 27 |
trust_remote_code=True,
|
| 28 |
-
use_flash_attention_2=False
|
| 29 |
).to(torch.float32)
|
| 30 |
|
| 31 |
-
|
| 32 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 33 |
model_path,
|
| 34 |
bos_token="</s>",
|
|
@@ -40,23 +41,41 @@ tokenizer = AutoTokenizer.from_pretrained(
|
|
| 40 |
clean_up_tokenization_spaces=False # Or set to True if you prefer
|
| 41 |
)
|
| 42 |
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
-
|
|
|
|
| 61 |
|
| 62 |
-
|
|
|
|
| 1 |
import torch
|
| 2 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 3 |
import time
|
| 4 |
+
import gradio as gr
|
| 5 |
|
| 6 |
def generate_prompt(instruction, input=""):
|
| 7 |
instruction = instruction.strip().replace('\r\n','\n').replace('\n\n','\n')
|
|
|
|
| 26 |
model = AutoModelForCausalLM.from_pretrained(
|
| 27 |
model_path,
|
| 28 |
trust_remote_code=True,
|
| 29 |
+
use_flash_attention_2=False
|
| 30 |
).to(torch.float32)
|
| 31 |
|
| 32 |
+
# Create a custom tokenizer (make sure to download vocab.json)
|
| 33 |
tokenizer = AutoTokenizer.from_pretrained(
|
| 34 |
model_path,
|
| 35 |
bos_token="</s>",
|
|
|
|
| 41 |
clean_up_tokenization_spaces=False # Or set to True if you prefer
|
| 42 |
)
|
| 43 |
|
| 44 |
+
# Function to handle text generation with word-by-word output and stop sequence
|
| 45 |
+
def generate_text(input_text):
|
| 46 |
+
prompt = generate_prompt(input_text)
|
| 47 |
+
input_ids = tokenizer(prompt, return_tensors="pt").input_ids
|
| 48 |
+
|
| 49 |
+
generated_text = ""
|
| 50 |
+
stop_sequence_found = False
|
| 51 |
+
for i in range(333):
|
| 52 |
+
output = model.generate(input_ids, max_new_tokens=1, do_sample=True, temperature=1.0, top_p=0.3, top_k=0)
|
| 53 |
+
new_word = tokenizer.decode(output[0][-1:], skip_special_tokens=True)
|
| 54 |
+
|
| 55 |
+
print(new_word, end="", flush=True)
|
| 56 |
+
generated_text += new_word
|
| 57 |
+
|
| 58 |
+
if new_word == '\n' or new_word == '.':
|
| 59 |
+
stop_sequence_found = True
|
| 60 |
+
break
|
| 61 |
+
|
| 62 |
+
input_ids = output
|
| 63 |
+
|
| 64 |
+
if stop_sequence_found:
|
| 65 |
+
print("\n(Stop sequence found)")
|
| 66 |
+
print()
|
| 67 |
+
return generated_text
|
| 68 |
+
|
| 69 |
+
# Create the Gradio interface
|
| 70 |
+
iface = gr.Interface(
|
| 71 |
+
fn=generate_text,
|
| 72 |
+
inputs="text",
|
| 73 |
+
outputs="text",
|
| 74 |
+
title="RWKV Chatbot",
|
| 75 |
+
description="Enter your prompt below:",
|
| 76 |
+
)
|
| 77 |
|
| 78 |
+
# For local testing:
|
| 79 |
+
# iface.launch()
|
| 80 |
|
| 81 |
+
# Hugging Face Spaces will automatically launch the interface.
|