Commit-Graf

133 Incheckningar

Upphovsman SHA1 Meddelande Datum
devbadxyz 44565efdaa fix(chatterbox): disable MIOpen backend to avoid broken RNN kernel 2026-09-13 00:09:23 +02:00
devbadxyz c4d28a5edd fix(chatterbox): install chatterbox-tts with --no-deps to protect ROCm torch 2026-09-12 23:54:16 +02:00
devbadxyz b8de22def2 feat(talkbot): add Chatterbox TTS backend, generalize TTS env vars 2026-09-12 23:38:46 +02:00
devbadxyz 35c9fa14de fix(omnivoice): join internal network and fix GPU detection race 2026-09-12 23:25:56 +02:00
devbadxyz 64b586913d fix(qwentts): fix GPU detection race, make ROCm flash-attn build opt-in 2026-09-12 21:50:42 +02:00
devbadxyz c93f50460a fix(qwentts): fix GPU detection race and add ROCm flash-attn support 2026-09-12 21:28:24 +02:00
devbadxyz 6c859990b6 fix(qwentts): fix GPU detection race and add ROCm flash-attn support 2026-09-12 21:26:43 +02:00
devbadxyz 90ff24a041 fix(qwentts): fix GPU detection race and add ROCm flash-attn support 2026-09-12 21:24:56 +02:00
devbadxyz 0b202e6409 fix(qwentts): fix GPU detection race and add ROCm flash-attn support 2026-09-12 21:18:21 +02:00
badstorm 33f2d8278e feat(talkbot): split voice/text queues and trim verbose log output 2026-09-11 22:15:39 +02:00
badstorm 44214b6084 feat(talkbot): reprocess cited messages and raise pipeline timeouts 2026-09-11 21:54:50 +02:00
badstorm 47d1f10e68 feat: add OmniVoice TTS container, upgrade whisper model, tune qwentts tone 2026-09-10 20:43:33 +02:00
badstorm 7a03174b60 fix(talkbot): post translated audio even if the original can't be deleted 2026-09-10 18:33:15 +02:00
badstorm 8f3df2c48f feat: auto-detect qwentts device, add build script, tune TTS defaults, drop unsupported languages 2026-09-10 17:30:29 +02:00
badstorm d1fb392ba5 feat(talkbot): add optional llama.cpp API key and stop double-posting text after audio 2026-09-10 17:14:06 +02:00
devbadxyz f15443df7a feat(talkbot): preserve text messages and reply with translations 2026-09-05 17:09:30 +02:00
devbadxyz 5608628e71 fix(talkbot): reply to image messages with translated captions 2026-09-05 17:06:39 +02:00
devbadxyz 162416ae1a feat(talkbot): preserve image captions and support replies 2026-09-05 17:02:54 +02:00
devbadxyz 884c30b546 feat(talkbot): add webhook payload debug logging 2026-09-05 16:58:03 +02:00
devbadxyz 3bf0f8bee8 feat(talkbot): make translation output limit configurable 2026-09-05 16:22:51 +02:00
devbadxyz 93bb6ffb5a fix(talkbot): improve text translation language handling 2026-09-05 15:56:32 +02:00
devbadxyz 6062c76cf8 fix(talkbot): retry failed voice message processing 2026-09-05 15:18:42 +02:00
devbadxyz 11d64a600d fix(talkbot): convert incoming audio to WAV before transcription 2026-09-05 14:38:56 +02:00
devbadxyz f6150409f0 Merge branch 'main' of https://code.badstorm.xyz/SRV/bdi_podman_serverconf 2026-09-05 14:32:02 +02:00
badstorm 863df5e17b gitignore: exclude Python bytecode cache 2026-09-05 00:35:47 +02:00
badstorm 260ec51442 containers/talkbot: document NC_URL trusted_domains requirement 2026-09-05 00:30:42 +02:00
badstorm 1795b30f79 containers/talkbot: prefix translated messages with the sender's name 2026-09-05 00:24:46 +02:00
badstorm 513ac804bf containers/talkbot: fix LLAMACPP_URL and retry failed queue jobs 2026-09-05 00:17:07 +02:00
badstorm cd52d6612c containers/talkbot: serialize webhook processing through a persistent queue 2026-09-03 18:45:09 +02:00
badstorm f36bc031fe containers/talkbot: fall back to default timeout on empty env var 2026-09-03 17:55:56 +02:00
badstorm 6331d5a0aa Add talkbot, qwen-tts, sdcpp containers and CPU whisper build 2026-09-03 17:54:30 +02:00
devbadxyz 25537692c8 fix(somedl): install yt-dlp nightly to avoid YouTube 403 errors 2026-08-21 17:04:49 +02:00
devbadxyz 56721db96c Merge branch 'main' of https://code.badstorm.xyz/SRV/bdi_podman_serverconf 2026-08-21 16:08:47 +02:00
devbadxyz b187b3201e Update containers/odoo/odoo-entrypoint.sh 2026-08-17 14:41:37 +00:00
devbadxyz 70bba2d2c0 Rinominati i modelli nel preset 2026-08-09 00:00:41 +02:00
devbadxyz 7b29728ca7 Rimossa variabile LLAMA_CACHE 2026-08-08 23:28:18 +02:00
devbadxyz 9f53851fbc containers/llamacpp: expose only preset models by pointing LLAMA_CACHE empty 2026-08-08 21:20:27 +02:00
devbadxyz e68fdbdaad containers/llamacpp: stop router from downloading the baked-in single model 2026-08-08 21:12:50 +02:00
devbadxyz 6f9e5d6a9e containers/llamacpp: full preset router mode (multi-model + embedding), drop
dedicated embedding container
2026-08-08 20:35:03 +02:00
devbadxyz 4bdc545233 containers/llamacpp: full preset router mode (multi-model + embedding), drop
dedicated embedding container
2026-08-08 20:34:57 +02:00
devbadxyz 1c41832fd5 containers/llamacpp: switch to router mode with presets and add watchdog 2026-08-08 20:24:45 +02:00
devbadxyz 7ce98859f1 Update containers/llamacpp/llamacpp.container 2026-08-08 16:58:46 +00:00
devbadxyz b38ec0f8e2 Update containers/gitea/gitea.container 2026-08-08 13:24:06 +00:00
devbadxyz 29c29eb728 Replace spotDL with SomeDL, rename container to somedl 2026-08-03 22:59:14 +02:00
badstorm 56bb7b2e43 Force PyTorch's default NCCL process-group timeout in vLLM's ROCm image 2026-08-02 23:03:56 +02:00
badstorm 624e79aaa6 Standardize llama.cpp env vars and binary handling, add vLLM env passthrough 2026-08-02 18:53:14 +02:00
badstorm cded740792 Add single-node mode, tool/reasoning calling, and TheRock-based RCCL build for vLLM 2026-08-02 12:37:30 +02:00
badstorm 0145854ed5 Add single-node mode, tool/reasoning calling, and TheRock-based RCCL build for vLLM 2026-08-02 12:37:28 +02:00
badstorm 71a3a22d2e Fix builder image 2026-08-01 23:37:43 +02:00
badstorm d21a7780a2 Add vllm container 2026-08-01 23:33:51 +02:00