Spaces:
Running
Running
Update text_generator.py
Browse files- text_generator.py +106 -58
text_generator.py
CHANGED
@@ -1,68 +1,116 @@
|
|
1 |
-
import requests
|
2 |
import os
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
from transformers import Tool
|
7 |
-
# Import other necessary libraries if needed
|
8 |
|
9 |
class TextGenerationTool(Tool):
|
10 |
name = "text_generator"
|
11 |
-
description =
|
12 |
-
|
13 |
-
)
|
14 |
-
|
15 |
inputs = ["text"]
|
16 |
outputs = ["text"]
|
17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
def __call__(self, prompt: str):
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
#
|
25 |
-
|
26 |
-
|
27 |
-
# "inputs": "Can you please let us know more details about your ",
|
28 |
-
# }
|
29 |
|
30 |
-
#
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
#
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
|
|
|
|
|
|
|
|
|
|
51 |
|
52 |
-
# Define the payload for the request
|
53 |
-
#payload = {
|
54 |
-
# "inputs": prompt # Adjust this based on your model's input format
|
55 |
-
#}
|
56 |
-
|
57 |
-
# Make the request to the API
|
58 |
-
#generated_text = requests.post(API_URL, headers=headers, json=payload).json()
|
59 |
-
|
60 |
# Extract and return the generated text
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import os
|
2 |
+
import requests
|
3 |
+
import gradio as gr
|
4 |
+
from transformers import pipeline, Tool
|
|
|
|
|
5 |
|
6 |
class TextGenerationTool(Tool):
|
7 |
name = "text_generator"
|
8 |
+
description = "This is a tool for text generation. It takes a prompt as input and returns the generated text."
|
9 |
+
|
|
|
|
|
10 |
inputs = ["text"]
|
11 |
outputs = ["text"]
|
12 |
+
|
13 |
+
# Available text generation models
|
14 |
+
models = {
|
15 |
+
"orca": "microsoft/Orca-2-13b",
|
16 |
+
"gpt2-dolly": "lgaalves/gpt2-dolly",
|
17 |
+
"gpt2": "gpt2",
|
18 |
+
"bloom": "bigscience/bloom-560m",
|
19 |
+
"openchat": "openchat/openchat_3.5"
|
20 |
+
}
|
21 |
+
|
22 |
+
def __init__(self, default_model="orca", use_api=False):
|
23 |
+
"""Initialize with a default model and API preference."""
|
24 |
+
self.default_model = default_model
|
25 |
+
self.use_api = use_api
|
26 |
+
self._pipelines = {}
|
27 |
+
|
28 |
+
# Check for API token
|
29 |
+
self.token = os.environ.get('HF_token')
|
30 |
+
if self.token is None and use_api:
|
31 |
+
print("Warning: HF_token environment variable not set. API calls will fail.")
|
32 |
+
|
33 |
def __call__(self, prompt: str):
|
34 |
+
"""Process the input prompt and generate text."""
|
35 |
+
return self.generate_text(prompt)
|
36 |
+
|
37 |
+
def generate_text(self, prompt, model_key=None, max_length=500, temperature=0.7):
|
38 |
+
"""Generate text based on the prompt using the specified or default model."""
|
39 |
+
# Determine which model to use
|
40 |
+
model_key = model_key or self.default_model
|
41 |
+
model_name = self.models.get(model_key, self.models[self.default_model])
|
|
|
|
|
42 |
|
43 |
+
# Generate using API if specified
|
44 |
+
if self.use_api and model_key == "openchat":
|
45 |
+
return self._generate_via_api(prompt, model_name)
|
46 |
+
|
47 |
+
# Otherwise use local pipeline
|
48 |
+
return self._generate_via_pipeline(prompt, model_name, max_length, temperature)
|
49 |
+
|
50 |
+
def _generate_via_pipeline(self, prompt, model_name, max_length, temperature):
|
51 |
+
"""Generate text using a local pipeline."""
|
52 |
+
# Get or create the pipeline
|
53 |
+
if model_name not in self._pipelines:
|
54 |
+
self._pipelines[model_name] = pipeline(
|
55 |
+
"text-generation",
|
56 |
+
model=model_name,
|
57 |
+
token=self.token
|
58 |
+
)
|
59 |
+
|
60 |
+
generator = self._pipelines[model_name]
|
61 |
+
|
62 |
+
# Generate text
|
63 |
+
result = generator(
|
64 |
+
prompt,
|
65 |
+
max_length=max_length,
|
66 |
+
num_return_sequences=1,
|
67 |
+
temperature=temperature
|
68 |
+
)
|
69 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
# Extract and return the generated text
|
71 |
+
if isinstance(result, list) and len(result) > 0:
|
72 |
+
if isinstance(result[0], dict) and 'generated_text' in result[0]:
|
73 |
+
return result[0]['generated_text']
|
74 |
+
return result[0]
|
75 |
+
|
76 |
+
return str(result)
|
77 |
+
|
78 |
+
def _generate_via_api(self, prompt, model_name):
|
79 |
+
"""Generate text by calling the Hugging Face API."""
|
80 |
+
if not self.token:
|
81 |
+
return "Error: HF_token not set. Cannot use API."
|
82 |
+
|
83 |
+
api_url = f"https://api-inference.huggingface.co/models/{model_name}"
|
84 |
+
headers = {"Authorization": f"Bearer {self.token}"}
|
85 |
+
payload = {"inputs": prompt}
|
86 |
+
|
87 |
+
try:
|
88 |
+
response = requests.post(api_url, headers=headers, json=payload)
|
89 |
+
response.raise_for_status() # Raise exception for HTTP errors
|
90 |
+
|
91 |
+
result = response.json()
|
92 |
+
|
93 |
+
# Handle different response formats
|
94 |
+
if isinstance(result, list) and len(result) > 0:
|
95 |
+
if isinstance(result[0], dict) and 'generated_text' in result[0]:
|
96 |
+
return result[0]['generated_text']
|
97 |
+
elif isinstance(result, dict) and 'generated_text' in result:
|
98 |
+
return result['generated_text']
|
99 |
+
|
100 |
+
# Fall back to returning the raw response
|
101 |
+
return str(result)
|
102 |
+
|
103 |
+
except Exception as e:
|
104 |
+
return f"Error generating text: {str(e)}"
|
105 |
+
|
106 |
+
# For standalone testing
|
107 |
+
if __name__ == "__main__":
|
108 |
+
# Create an instance of the TextGenerationTool
|
109 |
+
text_generator = TextGenerationTool(default_model="gpt2")
|
110 |
+
|
111 |
+
# Test with a simple prompt
|
112 |
+
test_prompt = "Once upon a time in a digital world,"
|
113 |
+
result = text_generator(test_prompt)
|
114 |
+
|
115 |
+
print(f"Prompt: {test_prompt}")
|
116 |
+
print(f"Generated text:\n{result}")
|