Skip to content

fix: adjust default params to make more use the gpu #151

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 2 commits into
base: master
Choose a base branch
from
Draft

fix: adjust default params to make more use the gpu #151

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5bd8576
fix: adjust default params to fully use the gpu
kyteinsky Jan 31, 2025
238b348
update readme: add avx2 requirement
kyteinsky Jan 31, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,8 @@
> [!NOTE]
> Be mindful to install the backend before the Context Chat php app (Context Chat php app would sends all the user-accessible files to the backend for indexing in the background. It is not an issue even if the request fails to an uninitialised backend since those files would be tried again in the next background job run.)
>
> The CPU (or the virtual CPU) should support AVX2 instructions for the embedder/LLM to work.
>
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Needs to be added to the docs as well

> The HTTP request timeout is 50 minutes for all requests and can be changed with the `request_timeout` app config for the php app `context_chat` using the occ command (`occ config:app:set context_chat request_timeout --value=3000`, value is in seconds). The same also needs to be done for docker socket proxy. See [Slow responding ExApps](https://github.com/cloud-py-api/docker-socket-proxy?tab=readme-ov-file#slow-responding-exapps)
>
> An end-to-end example on how to build and register the backend manually (with CUDA) is at the end of this readme
Expand Down
5 changes: 3 additions & 2 deletions config.cpu.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,13 +20,14 @@ embedding:
port: 5000
workers: 1
offload_after_mins: 15 # in minutes
request_timeout: 1800 # in seconds
request_timeout: 1680 # in seconds
llama:
# 'model_alias' is reserved
# 'embedding' is always set to True
model: multilingual-e5-large-instruct-q6_k.gguf
n_batch: 16
n_batch: 512
n_ctx: 8192
logits_all: false

llm:
nc_texttotext:
Expand Down
5 changes: 3 additions & 2 deletions config.gpu.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,14 +20,15 @@ embedding:
port: 5000
workers: 1
offload_after_mins: 15 # in minutes
request_timeout: 1800 # in seconds
request_timeout: 1680 # in seconds
llama:
# 'model_alias' is reserved
# 'embedding' is always set to True
model: multilingual-e5-large-instruct-q6_k.gguf
n_batch: 16
n_batch: 512
n_ctx: 8192
n_gpu_layers: -1
logits_all: false

llm:
nc_texttotext:
Expand Down
Loading