Concedo
fb7e72352e
benchmark includes ver
2024-05-02 14:17:48 +08:00
Concedo
e9978bfac0
resize window dimensions
2024-05-01 17:38:49 +08:00
Concedo
c65448d17a
add flash attention toggle
2024-04-30 21:29:11 +08:00
Concedo
f6ab0f873f
gui benchmark
2024-04-28 11:42:26 +08:00
Concedo
c230b78906
refactored a lot of code, remove bantokens, move it to api
2024-04-27 17:57:13 +08:00
Concedo
4ec8a9c57b
expose stop reason in generation
2024-04-27 01:12:12 +08:00
Concedo
327682fb97
undeprecate openblas, since it seems like there are still some valid cases for it
2024-04-26 20:21:29 +08:00
Concedo
544c36f751
merge, deprecate openblas
2024-04-26 19:24:59 +08:00
Concedo
0871c7cbd1
Add additional debug info and increased ctx sizes, fixed a bug loading vulkan config
2024-04-25 23:07:37 +08:00
Concedo
15ed96c25a
added support for resizing the window
2024-04-24 19:35:35 +08:00
Concedo
65bf69d104
Merge branch 'upstream' into concedo_experimental
...
# Conflicts:
# .github/workflows/bench.yml
# .github/workflows/server.yml
# CMakeLists.txt
# Makefile
# README-sycl.md
# flake.lock
2024-04-23 22:37:42 +08:00
Concedo
f449da3ba9
add additional warning for skiplauncher
2024-04-23 18:26:16 +08:00
Concedo
b4d2031215
merged, added ability to render special tokens
2024-04-22 18:19:58 +08:00
Concedo
a7b79ed2d7
smart buffered stop sequence workaround for SSE streaming mode.
2024-04-22 15:22:44 +08:00
Concedo
22c49e6b1e
Merge branch 'upstream' into concedo_experimental
...
# Conflicts:
# CMakeLists.txt
# Makefile
# README-sycl.md
# README.md
# scripts/compare-commits.sh
# tests/test-backend-ops.cpp
2024-04-19 10:35:50 +08:00
Concedo
b0d796fb49
use different cublas binaries
2024-04-17 17:14:22 +08:00
Concedo
eb380b500f
detect if other programs already using port
2024-04-15 15:54:03 +08:00
Concedo
010b2a5a71
fixed bad handling for malformed requests
2024-04-14 20:37:08 +08:00
Concedo
cfbe8cffbd
updated lite, added extra check in pyinstaller cleanup
2024-04-14 00:22:21 +08:00
Concedo
ab68447ad3
upver
2024-04-10 21:36:51 +08:00
Concedo
d8fb97211f
chat completions forced append stop sequences if already existing
2024-04-10 19:25:30 +08:00
Concedo
d1bb126605
Merge branch 'upstream' into concedo
...
# Conflicts:
# README.md
# llama.cpp
# otherarch/sdcpp/SDCPP_LICENSE
# scripts/sync-ggml-am.sh
# scripts/sync-ggml.sh
2024-04-09 17:18:35 +08:00
Concedo
d54af7fa31
updated swagger json link fix
2024-04-09 14:55:27 +08:00
Concedo
2bc11e00df
fixed stop sequence overwriting
2024-04-08 21:37:47 +08:00
Concedo
1ee5f355d4
try fix some compile issues (+1 squashed commits)
...
Squashed commits:
[e920e76b] try fix some compile issues
2024-04-08 20:01:46 +08:00
Concedo
fc881b4deb
fixed ssl cert path set if empty, added some basic cleanup of old temp dirs
2024-04-08 18:06:53 +08:00
Concedo
aa5124439d
horde workers pause themselves if recent local usage is detected (+1 squashed commits)
...
Squashed commits:
[7ebb80bc] horde workers pause themselves if recent local usage is detected
2024-04-07 23:22:50 +08:00
Concedo
6166fdfde4
added support for OAI chat completions adapter file, added default stop sequences to prevent chat compl leakage
2024-04-07 10:35:20 +08:00
Concedo
743687020d
fixed img2img
2024-04-06 17:29:44 +08:00
Concedo
073a279e70
change reference from kobold horde to ai horde
2024-03-18 22:35:49 +08:00
Concedo
5410e4644a
symlink docs
2024-03-17 22:27:26 +08:00
Concedo
9342071f9c
don't print url for localhost if remote tunnel
2024-03-16 22:19:04 +08:00
Concedo
7968bdebbb
added more stats in perf
2024-03-16 16:53:48 +08:00
Concedo
93d3871056
Merge branch 'upstream' into concedo_experimental
...
# Conflicts:
# .github/workflows/build.yml
# CMakeLists.txt
# Makefile
# ggml-metal.m
2024-03-15 10:37:48 +08:00
Concedo
f20fb7d778
mmq defaults to disabled only if full offload is possible
2024-03-14 23:34:45 +08:00
Concedo
f3b7651102
added ignoremissing param
2024-03-14 13:46:42 +08:00
Concedo
ec5dea14d7
merged, try to fix metal build
2024-03-14 11:15:50 +08:00
Concedo
ba950716a9
Merge branch 'master' into concedo_experimental
...
# Conflicts:
# CMakeLists.txt
# Makefile
# Package.swift
# README.md
# build.zig
# llama.cpp
# tests/test-tokenizer-1-bpe.cpp
# tests/test-tokenizer-1-llama.cpp
2024-03-13 11:21:58 +08:00
Concedo
edb05e761f
Update some prints
2024-03-12 21:40:36 +08:00
Concedo
6c6ad93f01
added basic support for password protection (+2 squashed commit)
...
Squashed commit:
[ff91ca72] added basic support for password protection
[91b0b208] updated docs
2024-03-12 19:47:12 +08:00
Concedo
a69bc44e7a
edit colab (+1 squashed commits)
...
Squashed commits:
[c7ccb99d] update colab with llava
2024-03-12 15:24:53 +08:00
Concedo
9229ea664e
if no existing filepath, do not use cwd, use last path instead
2024-03-11 22:19:38 +08:00
Stefan Kapusniak
4dd1c2b81a
Improve launcher file dialog initial paths ( #740 )
...
- In the launcher, if an existing value is set for a file value (e.g.
Model), use that file's directory the initial directory when the
file dialog is opened with 'Browse'.
- In the launcher always set the intial directory for 'Load' to
cwd.
2024-03-11 22:05:46 +08:00
Concedo
d59ec68753
added interrogate endpoint (+1 squashed commits)
...
Squashed commits:
[7bf96261] added interrogate endpoint
2024-03-11 18:50:18 +08:00
Concedo
e4946b96ea
support llava with gpt4v openai endpoint
2024-03-11 17:36:10 +08:00
Concedo
d943c739a8
wip submitting of llava image to backend
2024-03-10 17:14:27 +08:00
Concedo
c08d7e5042
wip integration of llava
2024-03-10 11:18:47 +08:00
Concedo
ca19199bc8
prevent sd logging when in quiet mode (+1 squashed commits)
...
Squashed commits:
[a4a1cdd5] fixed type conversion
2024-03-09 16:37:51 +08:00
Concedo
4682918965
better resolution scaling, support non-square images (+1 squashed commits)
...
Squashed commits:
[6efee4f2] better resolution scaling
2024-03-09 12:38:24 +08:00
Concedo
b4ca54401d
Merge branch 'master' into concedo_experimental
...
# Conflicts:
# Makefile
# README.md
# ci/run.sh
# ggml-quants.c
# ggml.c
# grammars/json.gbnf
# grammars/json_arr.gbnf
# llama.cpp
# scripts/compare-llama-bench.py
2024-03-09 10:58:49 +08:00
