vfbd
aeed9bd8f7
Fix base fairseq dense models when using accelerate with a GPU
2022-07-01 20:16:39 -04:00
vfbd
048bd0ff3b
Add support for setting the RNG seed and full determinism
2022-06-28 13:21:05 -04:00
ebolam
edd6dd7cd7
Fix for saved breakmodel settings on custom models
...
Fix for unit tests with new disk breakmodel
2022-06-27 10:12:54 -04:00
Henk
46678931b2
Better sentence spacing
2022-06-26 20:27:21 +02:00
vfbd
b99d1449c9
Remove trailing whitespace from submissions
2022-06-26 13:15:55 -04:00
Henk
fa97d28cb3
Nerys V2 for United
2022-06-25 14:06:51 +02:00
Henk
9e7eb80db4
Nerys V2 part 2
2022-06-25 14:03:19 +02:00
Henk
ecc6ee9474
Nerys V2
2022-06-25 13:47:49 +02:00
henk717
10e85db89d
Merge pull request #162 from VE-FORBRYDERNE/whitespace-cleanup
...
Story whitespace cleanup
2022-06-25 13:36:03 +02:00
Henk
d3fce44095
Merge branch 'main' into united
2022-06-24 18:31:45 +02:00
Henk
8be0964427
AIDG Import Fix
2022-06-24 18:29:06 +02:00
vfbd
4b16600e49
Clean up whitespace at the end of actions when loading story
...
Specifically, we merge blank actions into the next action and we move
whitespace at the end of non-blank actions to the beginning of the next
action.
2022-06-24 12:03:35 -04:00
vfbd
3da885d408
GPT-NeoX HF model badwords fix
2022-06-23 15:02:43 -04:00
henk717
8098f4ec8f
Merge branch 'KoboldAI:main' into united
2022-06-23 17:20:48 +02:00
vfbd
0eb9f8a879
Account for lnheader in budget calculation
2022-06-22 19:16:24 -04:00
vfbd
53034ee533
Delete all torch tensors before loading model
2022-06-22 12:07:36 -04:00
vfbd
922394c68f
Don't blacklist </s> token in "s" newline mode
2022-06-22 11:23:03 -04:00
Gnome Ann
8c594c6869
Correct the padding token for GPT-NeoX
2022-06-21 19:37:43 -04:00
Gnome Ann
a7f667c34c
Use NeoX badwords when loading from HF GPT-NeoX model
2022-06-21 19:33:25 -04:00
Gnome Ann
8593bf339b
Another typo fix
2022-06-21 15:36:25 -04:00
Gnome Ann
7e0ded6b47
Typo fix
2022-06-21 15:12:55 -04:00
Gnome Ann
91643be10a
Change soft prompt implementation to a more universal one
2022-06-21 15:03:43 -04:00
Gnome Ann
0ea4fa9c87
Automatically calculate badwords and pad_token_id
2022-06-21 14:35:52 -04:00
Gnome Ann
6b172306f6
move_model_to_devices no longer crashes if you don't have accelerate
2022-06-21 13:15:46 -04:00
Gnome Ann
ff69e9fbfe
Put layers_module_names, module_names and named_buffers in utils.py
2022-06-20 17:17:42 -04:00
Gnome Ann
1620ac4148
Lazy loader needs to cache named buffers of layers in the disk cache
2022-06-20 17:08:52 -04:00
Gnome Ann
ab5ab79003
Set primary device to CPU if in CPU-only mode
2022-06-20 16:25:01 -04:00
Gnome Ann
bd7d7b41a1
Don't enable accelerate if no layers are in disk cache or GPUs
2022-06-20 16:21:44 -04:00
Gnome Ann
90fd8b1845
Disk cache support in CPU-only mode
2022-06-20 16:06:09 -04:00
Gnome Ann
af07d7a15f
Disk cache support for computers with at least one GPU
2022-06-20 14:49:54 -04:00
Gnome Ann
47a58a36b8
Add disk cache slider
2022-06-19 22:53:30 -04:00
Gnome Ann
4dd59e0a9d
Correct the type hint for lazy_load_callback
2022-06-19 17:17:41 -04:00
Gnome Ann
21de36c4b0
Lazy loader now moves all non-layer weights to primary device
2022-06-19 16:44:23 -04:00
Gnome Ann
26c319519e
Lazy loader now attempts to pin layers if accelerate is enabled
2022-06-19 16:35:23 -04:00
Gnome Ann
042cf3e560
Automatically support soft prompts for all transformers models
2022-06-19 13:11:58 -04:00
Gnome Ann
cc56718a7e
Fix lazy loader putting too many layers on CPU
2022-06-19 00:29:35 -04:00
Gnome Ann
1380eb0bb0
Disable lazy loader when using GPT-2
2022-06-18 23:54:11 -04:00
Gnome Ann
f9732eb143
Always enable breakmodel if accelerate is available
2022-06-18 23:46:09 -04:00
Gnome Ann
8b4efc5d0a
Use accelerate.dispatch_model() instead of breakmodel if possible
2022-06-18 23:41:36 -04:00
Gnome Ann
f7ffdd7b6b
Add more model querying utilities
2022-06-18 18:16:56 -04:00
Gnome Ann
e143963161
Merge branch 'united' into accelerate
2022-06-18 13:47:38 -04:00
henk717
b209cf9868
NS mode as default
...
Experimental change that makes NS the default, more and more models seem to be requiring this as megatron based models are getting traction, neither does this seem to break the original models (with the exception of a user not being able to use </s> in generated outputs, the extremely rare case someone would be effected by this they can manually switch the mode by editing their settings file).
If this breaks nothing ns will remain the default, however the n mode should remain a choice for those who need it. In case it does get reversed I have also added the bloom model type to the ns list since its models require this.
2022-06-18 19:46:16 +02:00
Gnome Ann
0eedc541c8
Merge branch 'main' into united-merge
2022-06-18 13:39:23 -04:00
Gnome Ann
5e71f7fe97
Use slow tokenizer if fast tokenizer is not available
2022-06-17 21:08:37 -04:00
Gnome Ann
f71bae254a
Fix OPT tokenization problems
2022-06-17 13:29:42 -04:00
ebolam
2964175d8b
Fix for flaskwebgui
2022-06-17 08:17:22 -04:00
Henk
f112fc3493
Initial flaskwebgui support
2022-06-17 13:49:03 +02:00
Gnome Ann
8bdf17f598
Lazy loader can now use accelerate's init_empty_weights()
2022-06-16 18:56:16 -04:00
Gnome Ann
5253cdcb36
Lazy loader no longer requires map file except when loading to TPU
2022-06-16 18:45:11 -04:00
Gnome Ann
96d3d397ab
Don't use fallback loading if we run out of memory during loading
2022-06-15 14:35:32 -04:00
