rixty
/
KoboldAI-Client
mirror of https://github.com/KoboldAI/KoboldAI-Client.git
1
1
Fork 0
Code Issues Projects Releases Wiki Activity

Use eventlet instead of gevent-websocket

This commit is contained in:
Gnome Ann 2021-12-13 17:19:04 -05:00
parent fb6762bc1a
commit 338d437ea3
8 changed files with 112 additions and 101 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

190
aiserver.py
View File

@ -6,7 +6,12 @@
#==================================================================#
# External packages
import eventlet
eventlet.monkey_patch()
import os
os.environ['EVENTLET_THREADPOOL_SIZE'] = '1'
from eventlet import tpool
from os import path, getcwd
import re
import tkinter as tk
@ -559,7 +564,7 @@ from flask import Flask, render_template, Response, request
from flask_socketio import SocketIO, emit
app = Flask(__name__)
app.config['SECRET KEY'] = 'secret!'
socketio = SocketIO(app)
socketio = SocketIO(app, async_method="eventlet")
print("{0}OK!{1}".format(colors.GREEN, colors.END))
# Start transformers and create pipeline
@ -990,8 +995,8 @@ def load_lua_scripts():
try:
vars.lua_koboldbridge.obliterate_multiverse()
vars.lua_koboldbridge.load_corescript("default.lua")
vars.lua_koboldbridge.load_userscripts(filenames, modulenames, descriptions)
tpool.execute(vars.lua_koboldbridge.load_corescript, "default.lua")
tpool.execute(vars.lua_koboldbridge.load_userscripts, filenames, modulenames, descriptions)
except lupa.LuaError as e:
vars.lua_koboldbridge.obliterate_multiverse()
if(vars.serverstarted):
@ -1293,7 +1298,7 @@ def lua_is_custommodel():
def execute_inmod():
vars.lua_logname = ...
try:
vars.lua_koboldbridge.execute_inmod()
tpool.execute(vars.lua_koboldbridge.execute_inmod)
except lupa.LuaError as e:
vars.lua_koboldbridge.obliterate_multiverse()
emit('from_server', {'cmd': 'errmsg', 'data': 'Lua script error, please check console.'}, broadcast=True)
@ -1307,7 +1312,7 @@ def execute_genmod():
def execute_outmod():
try:
vars.lua_koboldbridge.execute_outmod()
tpool.execute(vars.lua_koboldbridge.execute_outmod)
except lupa.LuaError as e:
vars.lua_koboldbridge.obliterate_multiverse()
emit('from_server', {'cmd': 'errmsg', 'data': 'Lua script error, please check console.'}, broadcast=True)
@ -1315,6 +1320,9 @@ def execute_outmod():
print("{0}{1}{2}".format(colors.RED, str(e).replace("\033", ""), colors.END), file=sys.stderr)
print("{0}{1}{2}".format(colors.YELLOW, "Lua engine stopped; please open 'Userscripts' and press Load to reinitialize scripts.", colors.END), file=sys.stderr)
set_aibusy(0)
if(vars.lua_koboldbridge.resend_settings_required):
vars.lua_koboldbridge.resend_settings_required = False
lua_resend_settings()
#==================================================================#
# Lua runtime startup
@ -1348,7 +1356,6 @@ bridged = {
"has_setting": lua_has_setting,
"get_setting": lua_get_setting,
"set_setting": lua_set_setting,
"resend_settings": lua_resend_settings,
"set_chunk": lua_set_chunk,
"get_modeltype": lua_get_modeltype,
"get_modelbackend": lua_get_modelbackend,
@ -2157,100 +2164,106 @@ def calcsubmit(txt):
#==================================================================#
# Send text to generator and deal with output
#==================================================================#
def _generate(txt, minimum, maximum, found_entries):
gen_in = tokenizer.encode(txt, return_tensors="pt", truncation=True).long()
if(vars.sp is not None):
soft_tokens = torch.arange(
model.config.vocab_size,
model.config.vocab_size + vars.sp.shape[0],
)
gen_in = torch.cat((soft_tokens[None], gen_in), dim=-1)
if(vars.hascuda and vars.usegpu):
gen_in = gen_in.to(0)
elif(vars.hascuda and vars.breakmodel):
gen_in = gen_in.to(breakmodel.primary_device)
else:
gen_in = gen_in.to('cpu')
model.kai_scanner_head_length = gen_in.shape[-1]
model.kai_scanner_excluded_world_info = found_entries
actions = vars.actions
if(vars.dynamicscan):
actions = actions.copy()
with torch.no_grad():
already_generated = 0
numseqs = vars.numseqs
while True:
genout = generator(
gen_in,
do_sample=True,
min_length=minimum,
max_length=maximum-already_generated,
repetition_penalty=vars.rep_pen,
bad_words_ids=vars.badwordsids,
use_cache=True,
num_return_sequences=numseqs
)
already_generated += len(genout[0]) - len(gen_in[0])
if(model.kai_scanner.halt or not model.kai_scanner.regeneration_required):
break
assert genout.ndim >= 2
assert genout.shape[0] == vars.numseqs
if(already_generated != vars.lua_koboldbridge.generated_cols):
raise RuntimeError("WI scanning error")
for r in range(vars.numseqs):
for c in range(already_generated):
assert vars.lua_koboldbridge.generated[r+1][c+1] is not None
genout[r][genout.shape[-1] - already_generated - c] = vars.lua_koboldbridge.generated[r+1][c+1]
encoded = []
for i in range(vars.numseqs):
txt = tokenizer.decode(genout[i, -already_generated:])
winfo, mem, anotetxt, _found_entries = calcsubmitbudgetheader(txt, force_use_txt=True)
found_entries[i].update(_found_entries)
txt, _, _ = calcsubmitbudget(len(actions), winfo, mem, anotetxt, actions)
encoded.append(tokenizer.encode(txt, return_tensors="pt", truncation=True)[0].long().to(genout.device))
max_length = len(max(encoded, key=len))
encoded = torch.stack(tuple(torch.nn.functional.pad(e, (max_length - len(e), 0), value=model.config.pad_token_id or model.config.eos_token_id) for e in encoded))
genout = torch.cat(
(
encoded,
genout[..., -already_generated:],
),
dim=-1
)
if(vars.sp is not None):
soft_tokens = torch.arange(
model.config.vocab_size,
model.config.vocab_size + vars.sp.shape[0],
device=genout.device,
)
genout = torch.cat((soft_tokens.tile(vars.numseqs, 1), genout), dim=-1)
diff = genout.shape[-1] - gen_in.shape[-1]
minimum += diff
maximum += diff
gen_in = genout
model.kai_scanner_head_length = encoded.shape[-1]
numseqs = 1
return genout, already_generated
def generate(txt, minimum, maximum, found_entries=None):
if(found_entries is None):
found_entries = set()
found_entries = tuple(found_entries.copy() for _ in range(vars.numseqs))
print("{0}Min:{1}, Max:{2}, Txt:{3}{4}".format(colors.YELLOW, minimum, maximum, txt, colors.END))
# Store context in memory to use it for comparison with generated content
vars.lastctx = txt
# Clear CUDA cache if using GPU
if(vars.hascuda and (vars.usegpu or vars.breakmodel)):
gc.collect()
torch.cuda.empty_cache()
# Submit input text to generator
try:
gen_in = tokenizer.encode(txt, return_tensors="pt", truncation=True).long()
if(vars.sp is not None):
soft_tokens = torch.arange(
model.config.vocab_size,
model.config.vocab_size + vars.sp.shape[0],
)
gen_in = torch.cat((soft_tokens[None], gen_in), dim=-1)
if(vars.hascuda and vars.usegpu):
gen_in = gen_in.to(0)
elif(vars.hascuda and vars.breakmodel):
gen_in = gen_in.to(breakmodel.primary_device)
else:
gen_in = gen_in.to('cpu')
model.kai_scanner_head_length = gen_in.shape[-1]
model.kai_scanner_excluded_world_info = found_entries
actions = vars.actions
if(vars.dynamicscan):
actions = actions.copy()
with torch.no_grad():
already_generated = 0
numseqs = vars.numseqs
while True:
genout = generator(
gen_in,
do_sample=True,
min_length=minimum,
max_length=maximum-already_generated,
repetition_penalty=vars.rep_pen,
bad_words_ids=vars.badwordsids,
use_cache=True,
num_return_sequences=numseqs
)
already_generated += len(genout[0]) - len(gen_in[0])
if(model.kai_scanner.halt or not model.kai_scanner.regeneration_required):
break
assert genout.ndim >= 2
assert genout.shape[0] == vars.numseqs
if(already_generated != vars.lua_koboldbridge.generated_cols):
raise RuntimeError("WI scanning error")
for r in range(vars.numseqs):
for c in range(already_generated):
assert vars.lua_koboldbridge.generated[r+1][c+1] is not None
genout[r][genout.shape[-1] - already_generated - c] = vars.lua_koboldbridge.generated[r+1][c+1]
encoded = []
for i in range(vars.numseqs):
txt = tokenizer.decode(genout[i, -already_generated:])
winfo, mem, anotetxt, _found_entries = calcsubmitbudgetheader(txt, force_use_txt=True)
found_entries[i].update(_found_entries)
txt, _, _ = calcsubmitbudget(len(actions), winfo, mem, anotetxt, actions)
encoded.append(tokenizer.encode(txt, return_tensors="pt", truncation=True)[0].long().to(genout.device))
max_length = len(max(encoded, key=len))
encoded = torch.stack(tuple(torch.nn.functional.pad(e, (max_length - len(e), 0), value=model.config.pad_token_id or model.config.eos_token_id) for e in encoded))
genout = torch.cat(
(
encoded,
genout[..., -already_generated:],
),
dim=-1
)
if(vars.sp is not None):
soft_tokens = torch.arange(
model.config.vocab_size,
model.config.vocab_size + vars.sp.shape[0],
device=genout.device,
)
genout = torch.cat((soft_tokens.tile(vars.numseqs, 1), genout), dim=-1)
diff = genout.shape[-1] - gen_in.shape[-1]
minimum += diff
maximum += diff
gen_in = genout
model.kai_scanner_head_length = encoded.shape[-1]
numseqs = 1
genout, already_generated = tpool.execute(_generate, txt, minimum, maximum, found_entries)
except Exception as e:
if(issubclass(type(e), lupa.LuaError)):
vars.lua_koboldbridge.obliterate_multiverse()
@ -2448,7 +2461,8 @@ def tpumtjgenerate(txt, minimum, maximum, found_entries=None):
dtype=np.uint32
)
genout = tpu_mtj_backend.infer(
genout = tpool.execute(
tpu_mtj_backend.infer,
txt,
gen_len = maximum-minimum+1,
temp=vars.temp,
@ -3813,4 +3827,4 @@ if __name__ == "__main__":
webbrowser.open_new('http://localhost:5000')
print("{0}\nServer started!\nYou may now connect with a browser at http://127.0.0.1:5000/{1}".format(colors.GREEN, colors.END))
vars.serverstarted = True
socketio.run(app)
socketio.run(app, port=5000)

3
bridge.lua
View File

@ -1679,9 +1679,6 @@ return function(_python, _bridged)
end
end
end
if koboldbridge.resend_settings_required then
bridged.resend_settings()
end
koboldbridge.userstate = nil
return r
end

4
environments/finetuneanon.yml
View File

@ -10,11 +10,11 @@ dependencies:
- cudatoolkit=11.1
- tensorflow-gpu
- python=3.8.*
- gevent-websocket
- eventlet
- pip
- git
- pip:
- git+https://github.com/finetuneanon/transformers@gpt-neo-localattention3-rp-b
- flask-cloudflared
- flask-ngrok
- lupa
- lupa==1.10

4
environments/huggingface.yml
View File

@ -11,10 +11,10 @@ dependencies:
- python=3.8.*
- cudatoolkit=11.1
- transformers
- gevent-websocket
- eventlet
- pip
- git
- pip:
- flask-cloudflared
- flask-ngrok
- lupa
- lupa==1.10

4
environments/rocm-finetune.yml
View File

@ -6,7 +6,7 @@ dependencies:
- colorama
- flask-socketio
- python=3.8.*
- gevent-websocket
- eventlet
- pip
- git
- pip:
@ -16,4 +16,4 @@ dependencies:
- flask-cloudflared
- git+https://github.com/finetuneanon/transformers@gpt-neo-localattention3-rp-b
- flask-ngrok
- lupa
- lupa==1.10

4
environments/rocm.yml
View File

@ -8,7 +8,7 @@ dependencies:
- colorama
- flask-socketio
- python=3.8.*
- gevent-websocket
- eventlet
- pip
- git
- pip:
@ -17,4 +17,4 @@ dependencies:
- torchvision==0.11.1
- flask-cloudflared
- flask-ngrok
- lupa
- lupa==1.10

2
static/application.js
View File

@ -1619,7 +1619,7 @@ $(document).ready(function(){
seqselcontents = $("#seqselcontents");
// Connect to SocketIO server
socket = io.connect(window.document.origin);
socket = io.connect(window.document.origin, {transports: ['websocket', 'polling']});
socket.on('from_server', function(msg) {
if(msg.cmd == "connected") {

2
templates/index.html
View File

@ -7,7 +7,7 @@
<script src="static/jquery-3.6.0.min.js"></script>
<script src="static/jquery-ui.sortable.min.js"></script>
<script src="static/socket.io.min.js"></script>
<script src="static/application.js?ver=1.16.4e"></script>
<script src="static/application.js?ver=1.16.4f"></script>
<script src="static/bootstrap.min.js"></script>
<script src="static/bootstrap-toggle.min.js"></script>
<script src="static/rangy-core.min.js"></script>