ebolam
5d23f10bf3
Fix for mobile touch
2023-09-04 15:31:30 -04:00
henk717
5c6ff8faa5
Merge pull request #450 from Disty0/united
...
Initial Intel ARC support with IPEX
2023-09-04 20:56:34 +02:00
Disty0
7cd23ed1a8
Fix return_xpu
2023-09-04 11:43:59 +03:00
Disty0
8151948698
Initial Intel ARC support with IPEX
2023-09-04 04:06:41 +03:00
Henk
af330f0445
Remove legnth cap
2023-09-04 01:23:31 +02:00
Henk
cc213e028d
Ignore unknown generation parameters
2023-09-04 00:41:39 +02:00
Henk
0ab11e1322
Allow unlimited stop sequences
2023-09-03 23:25:35 +02:00
henk717
ce32dddef3
Merge pull request #449 from LostRuins/concedo_united_PR
...
Updated Kobold Lite to v60
2023-09-03 16:09:06 +02:00
Concedo
966b6f8a2e
Updated Kobold Lite to v60
2023-09-03 11:40:28 +08:00
Henk
4c03402e33
API Example Polish and Fixes
2023-09-02 03:41:01 +02:00
henk717
d1430d703d
Merge pull request #448 from YellowRoseCx/patch-3
...
Create api_example.py
2023-09-02 03:39:25 +02:00
YellowRoseCx
cb577bd491
Create api_example.py
2023-09-01 20:38:58 -05:00
Henk
565ab8a38f
badwordids -> badwordsids typofix
2023-08-30 17:00:51 +02:00
henk717
1d32d16ba4
Merge branch 'KoboldAI:main' into united
2023-08-30 13:13:20 +02:00
Llama
fe53cb2f7e
Merge pull request #70 from pi6am/feat/exllama-unban-eos
...
Hook up use_default_badwordids in exllama
2023-08-29 23:17:37 -07:00
Llama
d6ed75f993
Hook up use_default_badwordids in exllama
...
Use the value of the use_default_badwordids setting to configure
bad_words_ids. Also add square brackets to bad_words_ids if the
use_default_badwordids setting is True. Fix an issue with
attempting to use the tokenizer too early, and fix an exception
populating Lua bridge data when zero tokens are generated, which
can now happen if use_default_badwordids is False and the first
token generated is EOS.
2023-08-29 23:08:51 -07:00
Llama
36f53cc915
Merge pull request #69 from pi6am/merge/united-exllama
...
Merge branch henk717/united into exllama
2023-08-29 19:54:14 -07:00
Llama
b5b0e3faea
Merge branch 'henk717/united' into merge/united-exllama
2023-08-29 19:48:44 -07:00
Henk
49fa63052f
Allow EOS unbanning
2023-08-29 20:51:09 +02:00
Henk
d77acf17eb
Transformers 4.32.1
2023-08-28 22:55:33 +02:00
Henk
ec4cecf2f8
Restore Horde UI for local users
2023-08-28 21:46:30 +02:00
Henk
c29a5019b3
Disable Horde UI due to lockups
2023-08-28 20:02:29 +02:00
Llama
6e64763b63
Merge pull request #68 from pi6am/merge/united-exllama
...
Merge henk717/united into exllama
2023-08-28 09:57:35 -07:00
Llama
2c48e05f7c
Add exllama dependency back to requirements.
2023-08-28 09:52:31 -07:00
Llama
6151cbd053
Merge branch 'united' into merge/united-exllama
2023-08-28 09:32:19 -07:00
henk717
030ca26aa3
Merge pull request #445 from db0/scribe
...
fix: Avoid conflicting args
2023-08-28 18:04:15 +02:00
db0
b20b910568
remove debug
2023-08-28 17:49:34 +02:00
db0
31161409f6
fix argsv placement
2023-08-28 17:47:52 +02:00
db0
25b78abcd2
Merge branch 'united' into scribe
2023-08-28 17:45:49 +02:00
Llama
5229987ab7
Merge pull request #66 from pi6am/feat/exllama-config
...
Modify exllama to load unrenamed gptq quantized models
2023-08-28 00:09:50 -07:00
Llama
554af7b175
Modify exllama to load unrenamed gptq quantized models
...
Read config.json and enable exllama loading if the model has a
`quantization_config` with `quant_methdod` of `gptq`. Note that this
implementation is limited and only supports model.safetensors.
That said, this supports loading popular gptq quantized models
without renaming or symlinking the model file.
2023-08-27 23:56:02 -07:00
Llama
812df5ea56
Merge pull request #65 from pi6am/feat/exllama-badwords
...
Add the eos token to exllama bad words.
2023-08-27 17:03:25 -07:00
Llama
08ff7c138c
Add the eos token to exllama bad words.
...
The bos token was already hardcoded as a bad word id.
Store badwords in a list and iterate over them during generation.
Add the Llama eos token to the list of bad words.
Also support "single line mode", which adds newline (13) to badwords.
2023-08-27 16:34:52 -07:00
Henk
3e0b8279f2
Rename GPTQ loading
2023-08-27 20:51:14 +02:00
Llama
0d150e412e
Merge pull request #64 from pi6am/fix/multinomial-workaround
...
Resample to work around a bug in torch.multinomial
2023-08-26 22:42:21 -07:00
Llama
b7e38b4757
Resample to work around a bug in torch.multinomial
...
There is a bug in PyTorch 2.0.1 that allows torch.multinomial to
sometimes choose elements that have zero probability. Since
this is uncommon we can continue to use torch.multinomial as
long as we verify that the results are valid. If they aren't,
try again until the probability of each selected token is positive.
2023-08-26 22:26:26 -07:00
Henk
290f2ce05e
CPU only warning
2023-08-26 00:03:28 +02:00
db0
4b2d591354
avoid conflictinng sys args
2023-08-25 15:05:36 +02:00
Henk
f40236c04a
Modern llama tokenizer
2023-08-25 14:27:44 +02:00
Henk
2887467eec
Safetensors 0.3.3
2023-08-24 14:30:44 +02:00
henk717
d86f61151b
Working revision support
2023-08-23 22:07:37 +02:00
Henk
39c1b39b4a
Fix markers
2023-08-23 21:42:06 +02:00
Henk
5d9f180489
Fix typo
2023-08-23 21:36:26 +02:00
Henk
85810cd3fd
AutoGPTQ for Colab
2023-08-23 21:30:58 +02:00
Henk
c20ea949d7
Fix duplicate safetensors
2023-08-23 21:02:11 +02:00
Henk
91155ed2f3
HF dependencies
2023-08-23 20:34:40 +02:00
Llama
b1895de518
Merge pull request #63 from pi6am/feat/exllama-stoppers
...
Add stopper hooks suppport to exllama
2023-08-22 23:14:00 -07:00
Llama
b96d5d8646
Add stopper hooks suppport to exllama
2023-08-22 23:06:16 -07:00
Henk
f66173f2a0
Git gonna git
2023-08-22 20:43:44 +02:00
Henk
2b6dcbe55e
New Horde Worker
2023-08-22 20:40:04 +02:00
