Update app.py
Browse files
app.py
CHANGED
@@ -1,4 +1,3 @@
|
|
1 |
-
|
2 |
import gradio as gr
|
3 |
import gc, copy, re
|
4 |
import urllib.request
|
@@ -15,7 +14,7 @@ model = RWKV(model=title, strategy='cpu bf16')
|
|
15 |
pipeline = PIPELINE(model, "rwkv_vocab_v20230424")
|
16 |
|
17 |
def generate_prompt(instruction, input=None, history=None):
|
18 |
-
#
|
19 |
history_str = ""
|
20 |
for pair in history:
|
21 |
history_str += f"Instruction: {pair[0]}\n\nAssistant: {pair[1]}\n\n"
|
@@ -51,7 +50,7 @@ def evaluate(
|
|
51 |
top_p=0.5,
|
52 |
presencePenalty = 0.5,
|
53 |
countPenalty = 0.5,
|
54 |
-
history=None #
|
55 |
):
|
56 |
args = PIPELINE_ARGS(temperature = max(0.2, float(temperature)), top_p = float(top_p),
|
57 |
alpha_frequency = countPenalty,
|
@@ -61,7 +60,7 @@ def evaluate(
|
|
61 |
|
62 |
instruction = re.sub(r'\n{2,}', '\n', instruction).strip().replace('\r\n','\n')
|
63 |
input = re.sub(r'\n{2,}', '\n', input).strip().replace('\r\n','\n')
|
64 |
-
ctx = generate_prompt(instruction, input, history) #
|
65 |
print(ctx + "\n")
|
66 |
|
67 |
all_tokens = []
|
@@ -119,23 +118,64 @@ with gr.Blocks(title=title) as demo:
|
|
119 |
gr.Markdown(f"100% RNN RWKV-LM **trained on 100+ natural languages**. Demo limited to ctxlen {ctx_limit}. For best results, <b>keep your prompt short and clear</b>.")
|
120 |
with gr.Row():
|
121 |
with gr.Column():
|
122 |
-
instruction = gr.Textbox(lines=2, label="Instruction", value=
|
123 |
input = gr.Textbox(lines=2, label="Input", placeholder="")
|
124 |
token_count = gr.Slider(10, 512, label="Max Tokens", step=10, value=333)
|
125 |
temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=1.2)
|
126 |
top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.3)
|
127 |
presence_penalty = gr.Slider(0.0, 1.0, label="Presence Penalty", step=0.1, value=0)
|
128 |
count_penalty = gr.Slider(0.0, 1.0, label="Count Penalty", step=0.1, value=0.7)
|
|
|
|
|
|
|
|
|
|
|
129 |
data = gr.Dataset(components=[instruction, input, token_count, temperature, top_p, presence_penalty, count_penalty], samples=examples, label="Example Instructions", headers=["Instruction", "Input", "Max Tokens", "Temperature", "Top P", "Presence Penalty", "Count Penalty"])
|
|
|
|
|
130 |
data.click(lambda x: x, [data], [instruction, input, token_count, temperature, top_p, presence_penalty, count_penalty])
|
131 |
|
132 |
with gr.Tab("Chat mode"):
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
139 |
|
140 |
demo.queue(max_size=10)
|
141 |
demo.launch(share=False)
|
|
|
|
|
1 |
import gradio as gr
|
2 |
import gc, copy, re
|
3 |
import urllib.request
|
|
|
14 |
pipeline = PIPELINE(model, "rwkv_vocab_v20230424")
|
15 |
|
16 |
def generate_prompt(instruction, input=None, history=None):
|
17 |
+
# parse the chat history into a string of user and assistant messages
|
18 |
history_str = ""
|
19 |
for pair in history:
|
20 |
history_str += f"Instruction: {pair[0]}\n\nAssistant: {pair[1]}\n\n"
|
|
|
50 |
top_p=0.5,
|
51 |
presencePenalty = 0.5,
|
52 |
countPenalty = 0.5,
|
53 |
+
history=None # add the history parameter to the evaluate function
|
54 |
):
|
55 |
args = PIPELINE_ARGS(temperature = max(0.2, float(temperature)), top_p = float(top_p),
|
56 |
alpha_frequency = countPenalty,
|
|
|
60 |
|
61 |
instruction = re.sub(r'\n{2,}', '\n', instruction).strip().replace('\r\n','\n')
|
62 |
input = re.sub(r'\n{2,}', '\n', input).strip().replace('\r\n','\n')
|
63 |
+
ctx = generate_prompt(instruction, input, history) # pass the history to the generate_prompt function
|
64 |
print(ctx + "\n")
|
65 |
|
66 |
all_tokens = []
|
|
|
118 |
gr.Markdown(f"100% RNN RWKV-LM **trained on 100+ natural languages**. Demo limited to ctxlen {ctx_limit}. For best results, <b>keep your prompt short and clear</b>.")
|
119 |
with gr.Row():
|
120 |
with gr.Column():
|
121 |
+
instruction = gr.Textbox(lines=2, label="Instruction", value='東京で訪れるべき素晴らしい場所とその紹介をいくつか挙げてください。')
|
122 |
input = gr.Textbox(lines=2, label="Input", placeholder="")
|
123 |
token_count = gr.Slider(10, 512, label="Max Tokens", step=10, value=333)
|
124 |
temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=1.2)
|
125 |
top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.3)
|
126 |
presence_penalty = gr.Slider(0.0, 1.0, label="Presence Penalty", step=0.1, value=0)
|
127 |
count_penalty = gr.Slider(0.0, 1.0, label="Count Penalty", step=0.1, value=0.7)
|
128 |
+
with gr.Column():
|
129 |
+
with gr.Row():
|
130 |
+
submit = gr.Button("Submit", variant="primary")
|
131 |
+
clear = gr.Button("Clear", variant="secondary")
|
132 |
+
output = gr.Textbox(label="Output", lines=5)
|
133 |
data = gr.Dataset(components=[instruction, input, token_count, temperature, top_p, presence_penalty, count_penalty], samples=examples, label="Example Instructions", headers=["Instruction", "Input", "Max Tokens", "Temperature", "Top P", "Presence Penalty", "Count Penalty"])
|
134 |
+
submit.click(evaluate, [instruction, input, token_count, temperature, top_p, presence_penalty, count_penalty, []], [output])
|
135 |
+
clear.click(lambda: None, [], [output])
|
136 |
data.click(lambda x: x, [data], [instruction, input, token_count, temperature, top_p, presence_penalty, count_penalty])
|
137 |
|
138 |
with gr.Tab("Chat mode"):
|
139 |
+
with gr.Row():
|
140 |
+
chatbot = gr.Chatbot()
|
141 |
+
with gr.Column():
|
142 |
+
msg = gr.Textbox(scale=4, show_label=False, placeholder="Enter text and press enter", container=False)
|
143 |
+
clear = gr.Button("Clear")
|
144 |
+
with gr.Column():
|
145 |
+
token_count = gr.Slider(10, 512, label="Max Tokens", step=10, value=333)
|
146 |
+
temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=1.2)
|
147 |
+
top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.3)
|
148 |
+
presence_penalty = gr.Slider(0.0, 1.0, label="Presence Penalty", step=0.1, value=0)
|
149 |
+
count_penalty = gr.Slider(0.0, 1.0, label="Count Penalty", step=0.1, value=0.7)
|
150 |
+
|
151 |
+
def clear_chat():
|
152 |
+
return "", []
|
153 |
+
|
154 |
+
def user_msg(message, history):
|
155 |
+
history = history or []
|
156 |
+
return "", history + [[message, None]]
|
157 |
+
|
158 |
+
def chat(history):
|
159 |
+
# get the last user message and the additional parameters
|
160 |
+
message = history[-1][0]
|
161 |
+
instruction = msg.value
|
162 |
+
token_count = token_count.value
|
163 |
+
|
164 |
+
temperature = temperature.value
|
165 |
+
top_p = top_p.value
|
166 |
+
presence_penalty = presence_penalty.value
|
167 |
+
count_penalty = count_penalty.value
|
168 |
+
|
169 |
+
response = evaluate(instruction, None, token_count, temperature, top_p, presence_penalty, count_penalty, history)
|
170 |
+
|
171 |
+
history[-1][1] = response
|
172 |
+
return history
|
173 |
+
|
174 |
+
|
175 |
+
msg.submit(user_msg, [msg, chatbot], [msg, chatbot], queue=False).then(
|
176 |
+
chat, chatbot, chatbot, api_name="chat"
|
177 |
+
)
|
178 |
+
clear.click(clear_chat, None, [chatbot], queue=False)
|
179 |
|
180 |
demo.queue(max_size=10)
|
181 |
demo.launch(share=False)
|