jalr/tabbyAPI-ollama
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Tree: Switch to async generators

Browse source 
Async generation helps remove many roadblocks to managing tasks
using threads. It should allow for abortables and modern-day paradigms.

NOTE: Exllamav2 itself is not an asynchronous library. It's just
been added into tabby's async nature to allow for a fast and concurrent
API server. It's still being debated to run stream_ex in a separate
thread or manually manage it using asyncio.sleep(0)

Signed-off-by: kingbri <bdashore3@proton.me>
This commit is contained in:
kingbri 2024-03-14 10:27:39 -04:00 • committed by Brian Dashore
parent 33e2df50b7
commit 7fded4f183
10 changed files with 84 additions and 88 deletions
Download patch file Download diff file Expand all files Collapse all files

6
common/model.py
View file

@ -52,7 +52,7 @@ async def load_model_gen(model_path: pathlib.Path, **kwargs):
progress.start()
try:
for module, modules in load_status:
async for module, modules in load_status:
if module == 0:
loading_task = progress.add_task(
f"[cyan]Loading {model_type} modules", total=modules
@ -76,12 +76,12 @@ async def load_model(model_path: pathlib.Path, **kwargs):
pass
def load_loras(lora_dir, **kwargs):
async def load_loras(lora_dir, **kwargs):
"""Wrapper to load loras."""
if len(container.active_loras) > 0:
unload_loras()
return container.load_loras(lora_dir, **kwargs)
return await container.load_loras(lora_dir, **kwargs)
def unload_loras():