Synced repo using 'sync_with_huggingface' Github Action
Browse files- app.py +16 -1
- codellama.py +139 -0
app.py
CHANGED
@@ -9,7 +9,7 @@ from discord import app_commands
|
|
9 |
from discord.ext import commands
|
10 |
from falcon import continue_falcon, try_falcon
|
11 |
from musicgen import music_create
|
12 |
-
|
13 |
|
14 |
# HF GUILD SETTINGS
|
15 |
MY_GUILD_ID = 1077674588122648679 if os.getenv("TEST_ENV", False) else 879548962464493619
|
@@ -51,11 +51,26 @@ async def falcon(ctx, prompt: str):
|
|
51 |
print(f"Error: {e}")
|
52 |
|
53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
@client.event
|
55 |
async def on_message(message):
|
56 |
"""Checks channel and continues Falcon conversation if it's the right Discord Thread"""
|
57 |
try:
|
58 |
await continue_falcon(message)
|
|
|
59 |
except Exception as e:
|
60 |
print(f"Error: {e}")
|
61 |
|
|
|
9 |
from discord.ext import commands
|
10 |
from falcon import continue_falcon, try_falcon
|
11 |
from musicgen import music_create
|
12 |
+
from codellama import continue_codellama, try_codellama
|
13 |
|
14 |
# HF GUILD SETTINGS
|
15 |
MY_GUILD_ID = 1077674588122648679 if os.getenv("TEST_ENV", False) else 879548962464493619
|
|
|
51 |
print(f"Error: {e}")
|
52 |
|
53 |
|
54 |
+
@client.hybrid_command(
|
55 |
+
name="codellama",
|
56 |
+
with_app_command=True,
|
57 |
+
description="Enter a prompt to generate code!",
|
58 |
+
)
|
59 |
+
@app_commands.guilds(MY_GUILD)
|
60 |
+
async def codellama(ctx, prompt: str):
|
61 |
+
"""Audioldm2 generation"""
|
62 |
+
try:
|
63 |
+
await try_codellama(ctx, prompt)
|
64 |
+
except Exception as e:
|
65 |
+
print(f"Error: (app.py){e}")
|
66 |
+
|
67 |
+
|
68 |
@client.event
|
69 |
async def on_message(message):
|
70 |
"""Checks channel and continues Falcon conversation if it's the right Discord Thread"""
|
71 |
try:
|
72 |
await continue_falcon(message)
|
73 |
+
await continue_codellama(message)
|
74 |
except Exception as e:
|
75 |
print(f"Error: {e}")
|
76 |
|
codellama.py
ADDED
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import asyncio
|
2 |
+
import json
|
3 |
+
import os
|
4 |
+
|
5 |
+
from gradio_client import Client
|
6 |
+
|
7 |
+
HF_TOKEN = os.getenv("HF_TOKEN")
|
8 |
+
|
9 |
+
codellama = Client("https://huggingface-projects-codellama-13b-chat.hf.space/", HF_TOKEN)
|
10 |
+
|
11 |
+
BOT_USER_ID = 1102236653545861151 # real
|
12 |
+
CODELLAMA_CHANNEL_ID = 1147210106321256508 # real
|
13 |
+
|
14 |
+
|
15 |
+
codellama_threadid_userid_dictionary = {}
|
16 |
+
codellama_threadid_conversation = {}
|
17 |
+
|
18 |
+
|
19 |
+
def codellama_initial_generation(prompt, thread):
|
20 |
+
"""job.submit inside of run_in_executor = more consistent bot behavior"""
|
21 |
+
global codellama_threadid_conversation
|
22 |
+
|
23 |
+
chat_history = f"{thread.id}.json"
|
24 |
+
conversation = []
|
25 |
+
with open(chat_history, "w") as json_file:
|
26 |
+
json.dump(conversation, json_file)
|
27 |
+
|
28 |
+
job = codellama.submit(prompt, chat_history, fn_index=0)
|
29 |
+
|
30 |
+
while job.done() is False:
|
31 |
+
pass
|
32 |
+
else:
|
33 |
+
result = job.outputs()[-1]
|
34 |
+
with open(result, "r") as json_file:
|
35 |
+
data = json.load(json_file)
|
36 |
+
response = data[-1][-1]
|
37 |
+
conversation.append((prompt, response))
|
38 |
+
with open(chat_history, "w") as json_file:
|
39 |
+
json.dump(conversation, json_file)
|
40 |
+
|
41 |
+
codellama_threadid_conversation[thread.id] = chat_history
|
42 |
+
if len(response) > 1300:
|
43 |
+
response = response[:1300] + "...\nTruncating response due to discord api limits."
|
44 |
+
return response
|
45 |
+
|
46 |
+
|
47 |
+
async def try_codellama(ctx, prompt):
|
48 |
+
"""Generates text based on a given prompt"""
|
49 |
+
try:
|
50 |
+
global codellama_threadid_userid_dictionary # tracks userid-thread existence
|
51 |
+
global codellama_threadid_conversation
|
52 |
+
|
53 |
+
if ctx.author.id != BOT_USER_ID:
|
54 |
+
if ctx.channel.id == CODELLAMA_CHANNEL_ID:
|
55 |
+
message = await ctx.send(f"**{prompt}** - {ctx.author.mention}")
|
56 |
+
if len(prompt) > 99:
|
57 |
+
small_prompt = prompt[:99]
|
58 |
+
else:
|
59 |
+
small_prompt = prompt
|
60 |
+
thread = await message.create_thread(name=small_prompt, auto_archive_duration=60)
|
61 |
+
|
62 |
+
loop = asyncio.get_running_loop()
|
63 |
+
output_code = await loop.run_in_executor(None, codellama_initial_generation, prompt, thread)
|
64 |
+
codellama_threadid_userid_dictionary[thread.id] = ctx.author.id
|
65 |
+
|
66 |
+
print(output_code)
|
67 |
+
await thread.send(output_code)
|
68 |
+
except Exception as e:
|
69 |
+
print(f"try_codellama Error: {e}")
|
70 |
+
await ctx.send(f"Error: {e} <@811235357663297546> (try_codellama error)")
|
71 |
+
|
72 |
+
|
73 |
+
async def continue_codellama(message):
|
74 |
+
"""Continues a given conversation based on chat_history"""
|
75 |
+
try:
|
76 |
+
if not message.author.bot:
|
77 |
+
global codellama_threadid_userid_dictionary # tracks userid-thread existence
|
78 |
+
if message.channel.id in codellama_threadid_userid_dictionary: # is this a valid thread?
|
79 |
+
if codellama_threadid_userid_dictionary[message.channel.id] == message.author.id:
|
80 |
+
print("Safetychecks passed for continue_codellama")
|
81 |
+
global codellama_threadid_conversation
|
82 |
+
|
83 |
+
prompt = message.content
|
84 |
+
chat_history = codellama_threadid_conversation[message.channel.id]
|
85 |
+
|
86 |
+
# Check to see if conversation is ongoing or ended (>15000 characters)
|
87 |
+
with open(chat_history, "r") as json_file:
|
88 |
+
conversation = json.load(json_file)
|
89 |
+
total_characters = 0
|
90 |
+
for item in conversation:
|
91 |
+
for string in item:
|
92 |
+
total_characters += len(string)
|
93 |
+
|
94 |
+
if total_characters < 15000:
|
95 |
+
if os.environ.get("TEST_ENV") == "True":
|
96 |
+
print("Running codellama.submit")
|
97 |
+
job = codellama.submit(prompt, chat_history, fn_index=0)
|
98 |
+
while job.done() is False:
|
99 |
+
pass
|
100 |
+
else:
|
101 |
+
if os.environ.get("TEST_ENV") == "True":
|
102 |
+
print("Continue_codellama job done")
|
103 |
+
|
104 |
+
result = job.outputs()[-1]
|
105 |
+
with open(result, "r") as json_file:
|
106 |
+
data = json.load(json_file)
|
107 |
+
response = data[-1][-1]
|
108 |
+
|
109 |
+
with open(chat_history, "r") as json_file:
|
110 |
+
conversation = json.load(json_file)
|
111 |
+
|
112 |
+
conversation.append((prompt, response))
|
113 |
+
# now we have prompt, response, and the newly updated full conversation
|
114 |
+
|
115 |
+
with open(chat_history, "w") as json_file:
|
116 |
+
json.dump(conversation, json_file)
|
117 |
+
if os.environ.get("TEST_ENV") == "True":
|
118 |
+
print(prompt)
|
119 |
+
print(response)
|
120 |
+
print(conversation)
|
121 |
+
print(chat_history)
|
122 |
+
|
123 |
+
codellama_threadid_conversation[message.channel.id] = chat_history
|
124 |
+
if len(response) > 1300:
|
125 |
+
response = response[:1300] + "...\nTruncating response due to discord api limits."
|
126 |
+
|
127 |
+
await message.reply(response)
|
128 |
+
|
129 |
+
total_characters = 0
|
130 |
+
for item in conversation:
|
131 |
+
for string in item:
|
132 |
+
total_characters += len(string)
|
133 |
+
|
134 |
+
if total_characters >= 15000:
|
135 |
+
await message.reply("Conversation ending due to length, feel free to start a new one!")
|
136 |
+
|
137 |
+
except Exception as e:
|
138 |
+
print(f"continue_codellama Error: {e}")
|
139 |
+
await message.reply(f"Error: {e} <@811235357663297546> (continue_codellama error)")
|