Update app.py
Browse files
app.py
CHANGED
|
@@ -35,7 +35,7 @@ def initialize_history(chunks):
|
|
| 35 |
# Initialize history with initial chunks
|
| 36 |
history = initialize_history(all_chunks[:2]) # Starting with the first two chunks for example
|
| 37 |
|
| 38 |
-
def format_prompt_mixtral(message, history
|
| 39 |
prompt = "<s>"
|
| 40 |
prompt += f"{system_prompt_text}\n\n" # Add the system prompt
|
| 41 |
|
|
@@ -47,7 +47,7 @@ def format_prompt_mixtral(message, history, chunks):
|
|
| 47 |
prompt += f"[INST] {message} [/INST]"
|
| 48 |
return prompt
|
| 49 |
|
| 50 |
-
def chat_inf(message, history,
|
| 51 |
generate_kwargs = dict(
|
| 52 |
temperature=temp,
|
| 53 |
max_new_tokens=tokens,
|
|
@@ -57,7 +57,7 @@ def chat_inf(message, history, chunks, seed, temp, tokens, top_p, rep_p):
|
|
| 57 |
seed=seed,
|
| 58 |
)
|
| 59 |
|
| 60 |
-
formatted_prompt = format_prompt_mixtral(message, history
|
| 61 |
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
| 62 |
output = ""
|
| 63 |
for response in stream:
|
|
@@ -73,11 +73,11 @@ def clear_fn():
|
|
| 73 |
|
| 74 |
rand_val = random.randint(1, 1111111111111111)
|
| 75 |
|
| 76 |
-
def check_rand(
|
| 77 |
-
if
|
| 78 |
-
return
|
| 79 |
else:
|
| 80 |
-
return
|
| 81 |
|
| 82 |
with gr.Blocks() as app:
|
| 83 |
gr.HTML("""<center><h1 style='font-size:xx-large;'>PTT Chatbot</h1><br><h3>running on Huggingface Inference </h3><br><h7>EXPERIMENTAL</center>""")
|
|
@@ -105,7 +105,13 @@ with gr.Blocks() as app:
|
|
| 105 |
|
| 106 |
hid1 = gr.Number(value=1, visible=False)
|
| 107 |
|
| 108 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
|
| 110 |
stop_btn.click(None, None, None, cancels=[go])
|
| 111 |
clear_btn.click(clear_fn, None, [inp, chat])
|
|
|
|
| 35 |
# Initialize history with initial chunks
|
| 36 |
history = initialize_history(all_chunks[:2]) # Starting with the first two chunks for example
|
| 37 |
|
| 38 |
+
def format_prompt_mixtral(message, history):
|
| 39 |
prompt = "<s>"
|
| 40 |
prompt += f"{system_prompt_text}\n\n" # Add the system prompt
|
| 41 |
|
|
|
|
| 47 |
prompt += f"[INST] {message} [/INST]"
|
| 48 |
return prompt
|
| 49 |
|
| 50 |
+
def chat_inf(message, history, seed, temp, tokens, top_p, rep_p):
|
| 51 |
generate_kwargs = dict(
|
| 52 |
temperature=temp,
|
| 53 |
max_new_tokens=tokens,
|
|
|
|
| 57 |
seed=seed,
|
| 58 |
)
|
| 59 |
|
| 60 |
+
formatted_prompt = format_prompt_mixtral(message, history)
|
| 61 |
stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
|
| 62 |
output = ""
|
| 63 |
for response in stream:
|
|
|
|
| 73 |
|
| 74 |
rand_val = random.randint(1, 1111111111111111)
|
| 75 |
|
| 76 |
+
def check_rand(rand, val):
|
| 77 |
+
if rand:
|
| 78 |
+
return random.randint(1, 1111111111111111)
|
| 79 |
else:
|
| 80 |
+
return int(val)
|
| 81 |
|
| 82 |
with gr.Blocks() as app:
|
| 83 |
gr.HTML("""<center><h1 style='font-size:xx-large;'>PTT Chatbot</h1><br><h3>running on Huggingface Inference </h3><br><h7>EXPERIMENTAL</center>""")
|
|
|
|
| 105 |
|
| 106 |
hid1 = gr.Number(value=1, visible=False)
|
| 107 |
|
| 108 |
+
def on_chat(message, chat, seed, temp, tokens, top_p, rep_p):
|
| 109 |
+
chat.clear()
|
| 110 |
+
history = initialize_history(all_chunks[:2])
|
| 111 |
+
for response in chat_inf(message, history, seed, temp, tokens, top_p, rep_p):
|
| 112 |
+
chat.append(*response)
|
| 113 |
+
|
| 114 |
+
go = btn.click(on_chat, [inp, chat, seed, temp, tokens, top_p, rep_p], chat)
|
| 115 |
|
| 116 |
stop_btn.click(None, None, None, cancels=[go])
|
| 117 |
clear_btn.click(clear_fn, None, [inp, chat])
|