docs(server): clarify that --ctx-size is total context divided among parallel slots #17767
+5
−2
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Collaborator
|
this documentation is auto-generated, modify its source in |
Collaborator
When using --parallel N, the --ctx-size value is the TOTAL context divided among all slots, not the per-slot context. This is a common source of confusion (see ggml-org#11681, ggml-org#5732). Examples: - --ctx-size 4096 --parallel 4 → each slot gets 1024 tokens - To get 4096 tokens per slot with 4 parallel slots, use --ctx-size 16384 Updated the help text in arg.cpp (the source for auto-generated docs) for both --ctx-size and --parallel flags to clarify this behavior. Fixes ggml-org#11681
kitaekatt
force-pushed
the
docs/clarify-ctx-size-parallel
branch
from
b43b61d to
dc0d886
Compare
Author
|
I have moved the changes to arg.cpp! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
When using
--parallel N, the--ctx-sizevalue is the total context divided among all slots, not the per-slot context. This is a common source of confusion (see #11681, #5732).Changes
Added clarification to two flags in
tools/server/README.md:--ctx-size: Added note explaining that when using--parallel N, this is the total context divided among all slots. Each slot getsctx-size / paralleltokens. To allocate X tokens per slot with N parallel slots, set--ctx-sizetoX * N.--parallel: Added note that the total context is divided equally among these slots.Example
--ctx-size 4096 --parallel 4→ each slot gets 1024 tokens--ctx-size 16384 --parallel 4Related Issues
--ctx-sizeis divided by--paralleland cannot be increased? #11681 (ctx-size divided by parallel confusion)