Commit Graph

4221 Commits

Author SHA1 Message Date
0cc4m
e8741a1b57 Disable scaled_dot_product_attention if torch version < 2 2023-06-20 09:19:43 +02:00
0cc4m
a191855b37 Track token generation progress 2023-06-19 19:14:26 +02:00
0cc4m
e874f0c1c2 Add token streaming support for exllama 2023-06-19 19:14:26 +02:00
0cc4m
0c7eaefb1a Fix AMD ROCm exllama inference 2023-06-13 10:11:29 +02:00
0cc4m
ebf7e2cf57 Update GPTQ module to 0.0.6 2023-06-12 08:27:30 +02:00
0cc4m
0001ae00ab Add v2 with bias support (e.g. for Tulu-30b) 2023-06-12 07:18:22 +02:00
0cc4m
12df8220fb Add gpt_bigcode support, fix 8-bit GPTQ incoherence 2023-06-12 07:14:36 +02:00
0cc4m
47b371b9d3 Fix multigpu 2023-06-06 19:51:38 +02:00
0cc4m
39dfb18455 Replace exllama samplers with kobold's inbuilt ones 2023-06-06 19:21:34 +02:00
0cc4m
94520d5c80 Fix exllama model unload 2023-06-05 18:43:57 +02:00
0cc4m
b35f61e987 Basic exllama plugin 2023-06-04 15:40:12 +02:00
0cc4m
c82625490a Rename gptq backend folder 2023-06-04 12:31:24 +02:00
0cc4m
05a0bfe6c4 Don't show HF support if no HF model files are found 2023-06-04 09:44:28 +02:00
0cc4m
eace95cc72 Pull upstream changes, fix conflict 2023-06-04 09:06:31 +02:00
0cc4m
974328ed22 Add 4bit requirements to requirements.txt 2023-06-04 09:02:15 +02:00
0cc4m
b7838c7dde Fall back to autogptq if available and model not supported by gptq-koboldai 2023-06-04 08:06:48 +02:00
henk717
0ac2800601 Merge pull request #370 from Zurnaz/tpu_bfloat16
fix: tpu support models saved in bfloat16
2023-06-03 14:06:21 +02:00
Bogdan Drema
18dc706944 fix: tpu support models saved in bfloat16 2023-06-03 12:21:52 +01:00
henk717
8008a1c0b0 Merge pull request #369 from ebolam/Model_Plugins
Fix for --nobreakmodel forcing CPU
2023-06-02 19:35:05 +02:00
ebolam
5c4d580aac Fix for --nobreakmodel forcing CPU
Put importing of colab packages into a if function so it doesn't error out
2023-06-02 12:58:59 -04:00
henk717
856f16c0b2 Merge pull request #368 from LightSaveUs/united
Reverting the RTE
2023-06-02 14:45:04 +02:00
Henk
acfcde765b Allow breakmodel on Colab 2023-06-02 01:26:40 +02:00
LightSaveUs
e2c422aef8 Reverting the RTE (6B)
VE was right, concedo is right
2023-06-01 17:43:44 +03:00
LightSaveUs
077d64f6df Reverting the RTE (13B)
VE was right, concedo is right
2023-06-01 17:42:35 +03:00
LightSaveUs
cdf4098c97 Reverting the RTE (Custom)
VE was right, concedo is right
2023-06-01 17:40:42 +03:00
Henk
7d84466d9d Clean-up trust remote code 2023-05-31 22:48:29 +02:00
Henk
bda8c931f8 Horde backend fix 2023-05-31 16:56:08 +02:00
Henk
953f615e3d Conda deactivation fix 2023-05-31 02:29:26 +02:00
0cc4m
cf886de18b Remove leftover values fro koboldai_settings.py 2023-05-30 19:15:20 +02:00
Henk
c1abbf1473 More aggressive isolation 2023-05-30 12:56:22 +02:00
0cc4m
f1171dd3c1 Merge remote-tracking branch 'origin/united' into 4bit-plugin 2023-05-29 07:51:01 +02:00
henk717
bcee7113e5 Merge pull request #366 from ebolam/Model_Plugins
Anti-malware, nobreakmodel, and css fix
2023-05-29 00:43:54 +02:00
0cc4m
e49d35afc9 Add 4bit plugin 2023-05-28 22:54:36 +02:00
0cc4m
d71a63fa49 Merge ebolam's model-plugins branch 2023-05-28 09:26:13 +02:00
ebolam
f1d0be3a87 Fix for missing CSS in UI1. Added Malware blocker for model backends 2023-05-27 13:06:08 -04:00
ebolam
06c50f2d1e Merge branch 'Model_Plugins' of https://github.com/ebolam/KoboldAI into Model_Plugins 2023-05-27 09:42:30 -04:00
ebolam
6c8099381f Changed nobreakmodel to go to all model backends 2023-05-27 09:37:15 -04:00
henk717
97d2a78899 Merge pull request #362 from ebolam/Model_Plugins
Implement modular model backends Phase 1
2023-05-27 15:33:20 +02:00
ebolam
47276c3424 Bug Fix 2023-05-27 08:49:21 -04:00
Henk
9e7b813c8a Newest Horde Bridge 2023-05-27 14:19:47 +02:00
ebolam
cce5c1932c Fix for custom model names 2023-05-26 21:40:39 -04:00
ebolam
9bc9021843 Added better help message for model_parameters in command line arguments 2023-05-26 21:16:54 -04:00
ebolam
9723154bed Fix for --path 2023-05-26 20:10:11 -04:00
ebolam
acf5b40cd8 Bug fix 2023-05-26 19:38:37 -04:00
ebolam
64ef8ca7c2 Fix for UI1 not highlighting selected model 2023-05-26 14:04:26 -04:00
ebolam
51cea7eb9d Added ability to add labels that are based on validation data in model loading settings 2023-05-26 12:43:17 -04:00
ebolam
0376ab5715 KoboldAI API model name fix 2023-05-26 11:42:29 -04:00
ebolam
52f5d87906 Fix horde tokenizer 2023-05-26 11:25:28 -04:00
ebolam
2c82e9c5e0 GooseAI Fixes 2023-05-26 11:08:30 -04:00
ebolam
d2c95bc60f Fix for non-jailed menu path navigation 2023-05-26 10:33:59 -04:00