Commit Graph

297 Commits

Author SHA1 Message Date
kingbri e445aeec14 Tabby: Fix model name return on error
Tabby's model API is always /v1/model/list, so return "None" if the
request fails since that means a model is most likely not loaded.

Signed-off-by: kingbri <bdashore3@proton.me>
2023-11-23 00:13:28 -05:00
Cohee df4ed389bf System prompt for Claude 2 2023-11-21 22:11:26 +02:00
Cohee 96b87641ca Add OpenAI Whisper API 2023-11-19 20:30:34 +02:00
Cohee 96caddfd71 Add koboldcpp as Text Completion source 2023-11-19 17:14:53 +02:00
Cohee 53c3fc16c1 Assorted SD fixes 2023-11-19 00:40:21 +02:00
LenAnderson e520a50de2 add autorun command line argument 2023-11-18 00:09:42 +00:00
kingbri 5f8c615981 Server: Fix model info logs
Don't use a formatted string.

Signed-off-by: kingbri <bdashore3@proton.me>
2023-11-17 15:01:13 -05:00
kingbri f7ed574d04 Tabby: Decouple combined conditionals
Done by request.

Signed-off-by: kingbri <bdashore3@proton.me>
2023-11-17 14:51:44 -05:00
kingbri 4cfa267b1b API Tokenizer: Add support for TabbyAPI
Use Tabby's /v1/token endpoints.

Signed-off-by: kingbri <bdashore3@proton.me>
2023-11-17 01:48:03 -05:00
kingbri f31b996cb5 Text Generation: Add TabbyAPI support
TabbyAPI is an exllamav2 only API server that aims to provide a simple
experience for loading and chatting with exl2 models.

SillyTavern currently doesn't have the ability to load and unload models,
so only add the OAI compatible completion endpoints.

The repository can be found here:
https://github.com/theroyallab/tabbyAPI

Signed-off-by: kingbri <bdashore3@proton.me>
2023-11-17 00:37:39 -05:00
Cohee 3fb26d3927 Add lazy loading of sentencepiece tokenizers 2023-11-15 19:39:55 +02:00
Cohee 5b5e42361a Fix chat backups saving with incorrect file extension 2023-11-15 01:06:27 +02:00
Cohee 5fe8f70eb1 #1345 Add API endpoint for web search interaction 2023-11-14 00:16:41 +02:00
Cohee 7afe9e6481 #1343 Add status check bypass 2023-11-12 13:23:46 +02:00
Cohee 4c0b3fb7ae Add null checks for OR pricing 2023-11-12 13:07:57 +02:00
Cohee abe4bb6d54 Add error handling for empty Palm response 2023-11-10 15:55:49 +02:00
Cohee 81fe9aa699 Fix updated tokenization via ooba API 2023-11-09 19:39:08 +02:00
Cohee 3780321ae6 Fix ooba model name displaying as turbo 2023-11-09 19:27:19 +02:00
Cohee 0e89bf90bc Use correct tokenizers for logit bias for Mistral and Llama models over OpenRouter 2023-11-09 01:03:54 +02:00
Cohee 76b8880972 Normalize response statuses 2023-11-08 23:20:55 +02:00
Cohee 480099ee97 Mancer will work in legacy API mode. Remove Soft Prompt mentions. 2023-11-08 18:16:47 +02:00
Cohee 4f1dbaa34b Remove /v1 from the ooba API URLs 2023-11-08 17:09:33 +02:00
Cohee 740f6548a2 Increase timeouts of OAI out of quota requests 2023-11-08 12:07:14 +02:00
Cohee e76c18c104 Legacy ooba API compatibility shim 2023-11-08 10:13:28 +02:00
Cohee 865256f5c0 Fix ooba tokenization via API. Fix requiring streaming URL to generate 2023-11-08 03:38:04 +02:00
Cohee b2629d9718 Refactor status checks and indicators. 2023-11-08 02:52:03 +02:00
Cohee 00a1eaab28 Add timeout for status check 2023-11-08 01:47:31 +02:00
Cohee 2c7b954a8d #1328 New API schema for ooba / mancer / aphrodite 2023-11-08 00:17:13 +02:00
Cohee 2d2ff5230c
Merge pull request #1318 from artisticMink/feature/bulk-edit-enhanced
Feature/bulk edit enhanced
2023-11-07 20:47:06 +02:00
Cohee 1896732f17 Add DALL-E to OpenAI plugin 2023-11-06 21:47:00 +02:00
artisticMink c584da2ea9 Accept both v1 and v2 spec for editcharacterattributes endpoint 2023-11-06 16:13:34 +01:00
artisticMink 42d838a0b3 Refactor TavernCardValidator error property 2023-11-06 16:13:34 +01:00
artisticMink 6eb6527d11 Prototyping 2023-11-06 16:13:34 +01:00
Cohee f10833a516 Add prompt format override for OpenRouter 2023-11-03 00:34:22 +02:00
RossAscends 75c6e74d41 support for min P on Kobold API 2023-11-02 14:53:57 +09:00
Cohee 922007ea25 #1294 Fix selective logic for embedded lorebooks 2023-10-28 12:28:03 +03:00
Cohee 24f406917d Add seed to Kobold APi 2023-10-26 21:22:00 +03:00
Cohee 51d2c9feba Add UI option for group gen strategy 2023-10-25 22:39:31 +03:00
Cohee 5dbe2ebf29 Add chat file backups 2023-10-24 22:09:55 +03:00
Cohee 18fa925855 Use named constant 2023-10-21 23:33:17 +03:00
Cohee ca73a3a7b7 Merge branch 'release' into staging 2023-10-21 15:06:19 +03:00
Cohee 008fcece04 Rewrite to sync readdir, add try-catch, filter out invalid files 2023-10-21 15:04:36 +03:00
Xrystal 1d38109dcf Use JSON instead of json5 2023-10-21 18:56:51 +08:00
Xrystal 6f0f420063 Promisify getallchatsofcharacter 2023-10-21 18:55:52 +08:00
Cohee b167eb9e22 Add raw token ids support to OAI logit bias. Fix token counting for turbo models 2023-10-19 13:37:08 +03:00
Cohee 6a1b230c7e #1226 Add chat-bound lorebooks 2023-10-16 23:03:42 +03:00
Cohee 8a3547ecd0 #1248 Fix a server crash on unexpected models response 2023-10-16 20:25:51 +03:00
majick 89d7a05ea7 Missed a spot 2023-10-14 20:08:45 -10:00
majick a9ebd2f80e Make v2 spec read logspam slightly actionable
The existing logspam doesn't even tell the user what char is failing, which means it's
just going to emit every startup, forever, grinding at the poor user's soul, never
letting them know *which* of their 352 cards is causing the spam... day after day after
day.  Until one day, the poor beleaguered user will emit the world's ugliest PR to at
least find the broken card in an attempt kill it.

Ask me how I know.
2023-10-14 17:41:23 -10:00
50h100a 6e78ec11fb Check for errors on the `stream_end` event 2023-10-10 18:04:28 -04:00