81251b073a
Implement downloadable tokenizers
...
Closes #2574 , #2754
2024-09-06 16:28:34 +00:00
5fc16a2474
New AI21 Jamba + tokenizer
2024-08-26 12:07:36 +03:00
62fed2ff6a
fix: pass model parameter for aphrodite tokenizer
2024-08-24 22:43:20 +00:00
b68969ee3b
Small update to enum display + consistency
2024-08-19 20:18:44 +02:00
62818055af
Get available tokenizers from UI
2024-08-19 15:51:11 +01:00
5f92b8a09e
Add slash command to set tokenizer
2024-08-19 00:39:15 +01:00
e707def7dd
Add tokenizer for Gemma/Gemini
2024-08-16 00:00:43 +03:00
b9857eb315
Added block entropy endpoints for chat and image
2024-08-08 00:08:03 -04:00
6bdbe7299f
Merge branch 'staging' into vllm-tc
2024-07-06 11:18:38 +03:00
cf56bfb6a9
Add 01.AI as a chat completion source
2024-06-28 00:51:09 +03:00
6c87dfe10d
Add a token counting endpoint for vLLM
2024-06-26 13:34:02 +00:00
66454bb711
Add reverse proxy support to Google MakerSuite to allow some Google MakerSuite URLs to no longer be hardcoded with domain names. ( #2307 )
...
* Add reverse proxy support to Google MakerSuite.
* Remove hardcoded URLs for some Google MakerSuite API calls.
* Don't send real key to alt.endpoint
* Fix for image captioning
* Fix key validation
* +fix key check for mistral
* Fix caption key validation
* Fix tokenization endpoint use
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
2024-05-24 21:38:29 +03:00
181b5aff97
Add Groq as chat completion source
2024-05-05 18:53:12 +03:00
7bfd666321
Add llama 3 tokenizer
2024-05-03 23:59:39 +03:00
2a2a63c52c
Add Perplexity as Chat Completion source
2024-04-20 00:09:38 +03:00
1c4bad35b2
#2085 Implement async token counting
2024-04-13 21:05:31 +03:00
2cdfda9d69
Actually use getCurrentDreamGenModelTokenizer
2024-03-08 08:40:03 +02:00
f962ad5c02
Add OpenRouter as a text completion source
2024-02-25 22:47:07 +02:00
10d78f9a25
Aphrodite tokenizer
2024-02-02 01:09:03 +02:00
60044c18a4
Implement Token Probabilities UI using logprobs
2024-01-25 18:34:46 -06:00
86d715cc16
Improve tokenizer detection
2024-01-05 16:17:06 +02:00
a39b6b31f4
Allow API tokenizers for textgen bias/bans
2024-01-05 16:07:05 +02:00
5734dbd17c
Add custom endpoint type
2023-12-20 18:29:03 +02:00
67dd52c21b
#1309 Ollama text completion backend
2023-12-19 16:38:11 +02:00
edd737e8bd
#371 Add llama.cpp inference server support
2023-12-18 22:38:28 +02:00
b0d9f14534
Re-add Together as a text completion source
2023-12-17 23:38:03 +02:00
7acb61ab68
use appropriate tokenizer with mistral completion source
2023-12-16 08:47:51 +10:00
2045e414d1
lint: format fix
2023-12-14 20:57:43 +02:00
3e82a7d439
tokenizer changes and fixes. + a toggle
2023-12-14 16:31:08 +10:00
e26159c00d
refactor and rework palm request to work with the 'content' format and added an endpoint for googles tokenizer
2023-12-14 15:49:50 +10:00
cebd6e9e0f
Add API token ids from KoboldCpp
2023-12-14 01:28:18 +02:00
9acef0fae6
Horde doesn't support API tokenizers
2023-12-10 16:21:06 +02:00
f54bf99006
Fix token ids not displaying in "API_CURRENT" mode for TextGen
2023-12-10 16:09:00 +02:00
6957d9e7cf
Fix display names of Best match tokenizers
2023-12-10 16:03:25 +02:00
6e5eea5dba
Unbreak previously selected API tokenizer in dropdown
2023-12-10 15:56:38 +02:00
55976e61a3
Fix tokenizer override
...
I searched for all users of tokenizers.API, but missed that the menu
converts the numerical select values directly to enum values. I've used
the special tokenizer value 98 to represent "the tokenizer API for
whichever backend we're currently using".
2023-12-09 23:57:21 -05:00
014416546c
Add padding once in getTokenCount
...
This means we don't have to pass the "padding" parameter into every
function so they can add the padding themselves--we can do it in just
one place instead.
2023-12-09 20:53:16 -05:00
2f2cd197cc
Clean up tokenizer API code
...
Store the URLs for each tokenizer's action in one place at the top of
the file, instead of in a bunch of switch-cases. The URLs for the
textgen and Kobold APIs don't change and hence don't need to be
function arguments.
2023-12-09 20:48:41 -05:00
09465fbb97
Inline most get(...)TokenizerParams calls
...
For everything except textgenerationwebui, these params are now simple
enough that it doesn't make sense for them to be in a separate function.
2023-12-09 20:35:11 -05:00
30502ac949
Split up Kobold and textgenerationwebui endpoints
...
The endpoint was one big if/else statement that did two entirely
different things depending on the value of main_api. It makes more sense
for those to be two separate endpoints.
2023-12-09 20:26:24 -05:00
7486ab3886
Separate textgen and Kobold tokenization APIs
...
They function differently and have different logic and API parameters,
so it makes sense to count them as two different APIs. Kobold's API
doesn't return tokens, so it can only be used to count them.
There's still a lot of duplicate code which I will clean up in the
following commits.
2023-12-09 20:24:56 -05:00
18177c147d
Separate remote and server tokenization code paths
...
This lets us remove extraneous API params from paths where they aren't
needed.
2023-12-09 20:08:48 -05:00
ddd73a204a
Remove "remote" language from tokenizer functions
...
We'll be making a distinction between tokenizing *on* the server itself,
and tokenizing via the server having the AI service do it. It makes more
sense to use the term "remote" for the latter.
2023-12-09 19:49:22 -05:00
8bad059a62
Rename /tokenize_via_api endpoint
...
No redirect for this since I don't expect any extensions to be calling this directly.
2023-12-09 19:29:24 -05:00
57bc95133e
Rename tokenizer routes
...
They're all under tokenizers/ now, and there are "count", "encode", and
"decode" endpoints. This forms a clearer hierarchy.
2023-12-04 10:17:43 -05:00
9c33ddbafc
Make textgen settings type checks more concise
2023-12-03 14:56:01 -05:00
047c897ead
Remove is[API] functions
...
Just use an equality comparison. It's a bit longer, but only because
"textgenerationwebui_settings" is a long identifier.
2023-12-03 14:56:01 -05:00
ba54e3dea0
Replaces is_[api] params with api_type param
...
These were 5 mutually-exclusive booleans, which can be replaced with one
param that takes on 5 values, one for each API type.
2023-12-03 14:56:01 -05:00
64a3564892
lint: Comma dangle
2023-12-02 22:06:57 +02:00
c63cd87cc0
lint: Require semicolons
2023-12-02 21:11:06 +02:00
Cohee
AlpinDale
Wolfsblvt
Eugén Cowie
Edward Kim
sasha0552
daiaji
khanon
based
valadaptive