use external inference
Browse files
README.md
CHANGED
@@ -7,7 +7,7 @@ sdk: gradio
|
|
7 |
sdk_version: 3.18.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
-
license:
|
11 |
---
|
12 |
|
13 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
7 |
sdk_version: 3.18.0
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
+
license: creativeml-openrail-m
|
11 |
---
|
12 |
|
13 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
app.py
CHANGED
@@ -1,38 +1,38 @@
|
|
1 |
-
import torch
|
2 |
import gradio as gr
|
3 |
-
from transformers import pipeline
|
4 |
import os
|
5 |
from mtranslate import translate
|
|
|
6 |
|
7 |
-
device = torch.cuda.current_device() if torch.cuda.is_available() else "cpu"
|
8 |
HF_AUTH_TOKEN = os.environ.get("HF_AUTH_TOKEN")
|
9 |
-
|
10 |
-
|
11 |
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
num_beams
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
else:
|
22 |
-
|
23 |
-
num_beams = 1
|
24 |
-
print(user_input, decoding_methods, do_sample, top_k, top_p, temperature, repetition_penalty, penalty_alpha)
|
25 |
-
prompt = f"User: {user_input}\nAssistant: "
|
26 |
-
generated_text = text_generation(f"{prompt}", min_length=50, max_length=200, num_return_sequences=1,
|
27 |
-
num_beams=num_beams, do_sample=do_sample, top_k=top_k, top_p=top_p,
|
28 |
-
temperature=temperature, repetition_penalty=repetition_penalty,
|
29 |
-
penalty_alpha=penalty_alpha)
|
30 |
-
answer = generated_text[0]["generated_text"]
|
31 |
-
answer_without_prompt = answer[len(prompt)+1:]
|
32 |
-
user_input_en = translate(user_input, "en", "id")
|
33 |
-
answer_without_prompt_en = translate(answer_without_prompt, "en", "id")
|
34 |
-
return [(f"{user_input}\n", None), (answer_without_prompt, "")], \
|
35 |
-
[(f"{user_input_en}\n", None), (answer_without_prompt_en, "")]
|
36 |
|
37 |
|
38 |
css = """
|
@@ -55,7 +55,7 @@ with gr.Blocks(css=css) as demo:
|
|
55 |
user_input = gr.inputs.Textbox(placeholder="",
|
56 |
label="Ask me something in Indonesian or English",
|
57 |
default="Bagaimana cara mendidik anak supaya tidak berbohong?")
|
58 |
-
|
59 |
default="Sampling", label="Decoding Method")
|
60 |
num_beams = gr.inputs.Slider(label="Number of beams for beam search",
|
61 |
default=1, minimum=1, maximum=10, step=1)
|
@@ -85,7 +85,7 @@ with gr.Blocks(css=css) as demo:
|
|
85 |
gr.Markdown("![visitor badge](https://visitor-badge.glitch.me/badge?page_id=cahya_indochat)")
|
86 |
|
87 |
button_generate_story.click(get_answer,
|
88 |
-
inputs=[user_input,
|
89 |
repetition_penalty, penalty_alpha],
|
90 |
outputs=[generated_answer, generated_answer_en])
|
91 |
|
|
|
|
|
1 |
import gradio as gr
|
|
|
2 |
import os
|
3 |
from mtranslate import translate
|
4 |
+
import requests
|
5 |
|
|
|
6 |
HF_AUTH_TOKEN = os.environ.get("HF_AUTH_TOKEN")
|
7 |
+
indochat_api = 'https://cahya-indonesian-whisperer.hf.space/api/indochat/v1'
|
8 |
+
indochat_api_auth_token = os.getenv("INDOCHAT_API_AUTH_TOKEN", "")
|
9 |
|
10 |
+
def get_answer(user_input, decoding_method, num_beams, top_k, top_p, temperature, repetition_penalty, penalty_alpha):
|
11 |
+
print(user_input, decoding_method, top_k, top_p, temperature, repetition_penalty, penalty_alpha)
|
12 |
+
headers = {'Authorization': 'Bearer ' + indochat_api_auth_token}
|
13 |
+
data = {
|
14 |
+
"text": user_input,
|
15 |
+
"min_length": len(user_input) + 50,
|
16 |
+
"max_length": 300,
|
17 |
+
"decoding_method": decoding_method,
|
18 |
+
"num_beams": num_beams,
|
19 |
+
"top_k": top_k,
|
20 |
+
"top_p": top_p,
|
21 |
+
"temperature": temperature,
|
22 |
+
"seed": -1,
|
23 |
+
"repetition_penalty": repetition_penalty,
|
24 |
+
"penalty_alpha": penalty_alpha
|
25 |
+
}
|
26 |
+
r = requests.post(indochat_api, headers=headers, data=data)
|
27 |
+
if r.status_code == 200:
|
28 |
+
result = r.json()
|
29 |
+
answer = result["generated_text"]
|
30 |
+
user_input_en = translate(user_input, "en", "id")
|
31 |
+
answer_en = translate(answer, "en", "id")
|
32 |
+
return [(f"{user_input}\n", None), (answer, "")], \
|
33 |
+
[(f"{user_input_en}\n", None), (answer_en, "")]
|
34 |
else:
|
35 |
+
return "Error: " + r.text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
36 |
|
37 |
|
38 |
css = """
|
|
|
55 |
user_input = gr.inputs.Textbox(placeholder="",
|
56 |
label="Ask me something in Indonesian or English",
|
57 |
default="Bagaimana cara mendidik anak supaya tidak berbohong?")
|
58 |
+
decoding_method = gr.inputs.Dropdown(["Beam Search", "Sampling", "Contrastive Search"],
|
59 |
default="Sampling", label="Decoding Method")
|
60 |
num_beams = gr.inputs.Slider(label="Number of beams for beam search",
|
61 |
default=1, minimum=1, maximum=10, step=1)
|
|
|
85 |
gr.Markdown("![visitor badge](https://visitor-badge.glitch.me/badge?page_id=cahya_indochat)")
|
86 |
|
87 |
button_generate_story.click(get_answer,
|
88 |
+
inputs=[user_input, decoding_method, num_beams, top_k, top_p, temperature,
|
89 |
repetition_penalty, penalty_alpha],
|
90 |
outputs=[generated_answer, generated_answer_en])
|
91 |
|