oobabooga/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2026-04-04 14:17:28 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Split by rows instead of layers for llama.cpp multi-gpu (#5435)

Browse source
This commit is contained in:
Forkoz 2024-02-05 02:36:40 +00:00 • committed by GitHub
parent 3df7e151f7
commit 2a45620c85
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 9 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

1
modules/shared.py
View file

@ -129,6 +129,7 @@ group.add_argument('--numa', action='store_true', help='Activate NUMA task alloc
group.add_argument('--logits_all', action='store_true', help='Needs to be set for perplexity evaluation to work. Otherwise, ignore it, as it makes prompt processing slower.')
group.add_argument('--no_offload_kqv', action='store_true', help='Do not offload the K, Q, V to the GPU. This saves VRAM but reduces the performance.')
group.add_argument('--cache-capacity', type=str, help='Maximum cache capacity (llama-cpp-python). Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.')
group.add_argument('--row_split', action='store_true', help='Split multi-gpu by row instead of layer. Faster on some cards.')
# ExLlamaV2
group = parser.add_argument_group('ExLlamaV2')