Commit Graph

  • d133ef1443
    Reduces cognitive complexity of chat module and improves string operations blackmambaza 2024-06-01 23:30:07 +0200
  • bd2a18e089
    Fix ui_model_menu.py Forkoz 2024-06-01 12:28:20 +0000
  • 776f320d28
    Fix loaders.py Forkoz 2024-06-01 12:27:30 +0000
  • ca6765bbe1
    Fix ui.py Forkoz 2024-06-01 12:26:56 +0000
  • 03c1a9bae1
    Fix shared.py Forkoz 2024-06-01 12:25:55 +0000
  • 7ce96dc25f A small fix Touch-Night 2024-06-01 11:24:41 +0800
  • 248d9cc647 Also, you can use llamacpp_hf to load multimodal model, however it's not working now, you can only load it but can't use it to do anything. Touch-Night 2024-05-31 01:03:51 +0800
  • 9b1da75ebd Now you can load multimodal models in llama.cpp Touch-Night 2024-05-31 00:40:00 +0800
  • 904c53dec7 Add pipelines folder to .gitignore Touch-Night 2024-05-30 01:09:57 +0800
  • e8d714d514 Remove instructblip-pipeline from the repo. Touch-Night 2024-05-30 01:07:44 +0800
  • fcdd9d06f1 Intergrated multimodal functionality Touch-Night 2024-05-30 01:03:54 +0800
  • 0957563429
    Merge fbf4d6996e into a363cdfca1 Eve 2024-05-28 21:21:36 +0300
  • 686e7a8344 Comment. Belladore 2024-05-28 12:39:04 +0300
  • 1e9cf3d284 Fix error when bos_token_id is None. Belladore 2024-05-28 12:30:42 +0300
  • 47c774e8a3 Add proper warning message for decode failure nclok1405 2024-05-28 00:54:54 +0900
  • f5ca611569 Fallback to latin on decode failure nclok1405 2024-05-28 00:45:06 +0900
  • a363cdfca1
    Fix missing bos token for some models (including Llama-3) (#6050) Belladore 2024-05-27 15:21:30 +0300
  • c5064d074c Minor changes/checks oobabooga 2024-05-27 05:17:56 -0700
  • 8df68b05e9 Remove MinPLogitsWarper (it's now a transformers built-in) oobabooga 2024-05-27 05:03:30 -0700
  • 918ff94f1d Change DRY max match length. Belladore 2024-05-26 23:22:55 +0300
  • 62e76edcab Rename variable. Belladore 2024-05-26 23:14:14 +0300
  • 3c51b955e1 Fix DRY crash on very repetitive inputs. Belladore 2024-05-25 18:19:19 +0300
  • 1005c3d219 Refactor DRY sampler. Belladore 2024-05-24 20:52:11 +0300
  • c059d9b650 Minor fix to prevent DRY sampler crash on very long repetitive inputs. Belladore 2024-05-24 17:58:33 +0300
  • 4f1e96b9e3 Downloader: Add --model-dir argument, respect --model-dir in the UI oobabooga 2024-05-23 20:42:46 -0700
  • d78159ccb3 Fix missing bos token (affecting many llama 3 exl2 quants and probably other models too). Belladore 2024-05-24 01:59:19 +0300
  • bb7e410a0c Improve DRY sampler performance 2 of 2 (introduce Z algorithm). Belladore 2024-05-23 02:44:29 +0300
  • d9fdb3db71 Add docs for image generation Artificiangel 2024-05-23 08:44:15 -0400
  • ee7d2c7406 missing requirements for transcription endpoint Artificiangel 2024-05-23 08:09:54 -0400
  • 432b070bde Fix transcriptions endpoint Artificiangel 2024-05-23 08:07:51 -0400
  • ad54d524f7 Revert "Fix stopping strings for llama-3 and phi (#6043)" oobabooga 2024-05-22 17:18:08 -0700
  • 7c03e4af4d Improve DRY sampler performance 1 of 2 (simple data type changes). Belladore 2024-05-23 00:54:36 +0300
  • 5499bc9bc8
    Fix stopping strings for llama-3 and phi (#6043) oobabooga 2024-05-22 13:53:59 -0300
  • bfb9d7fc01 Simplify oobabooga 2024-05-22 09:44:14 -0700
  • ce89dbc5ff Account for llama.cpp loader oobabooga 2024-05-22 09:42:33 -0700
  • ee6728640b Minor fixes oobabooga 2024-05-22 08:24:16 -0700
  • 0730c191da Fix stopping strings for llama-3 and phi oobabooga 2024-05-22 08:18:02 -0700
  • a4df993d41
    Update README.md Shixian Sheng 2024-05-22 07:32:09 -0400
  • 52d74bb0d8
    Update README.md Shixian Sheng 2024-05-22 07:27:36 -0400
  • 68c46970d5
    Update README.md Shixian Sheng 2024-05-22 07:27:01 -0400
  • 39db799bf1 Added UnicodeDecodeError workaround for modules/llamacpp_model.py nclok1405 2024-05-22 09:54:35 +0900
  • 8aaa0a6f4e
    Fixed minor typo in docs - Training Tab.md (#6038) rohitanshu 2024-05-21 23:22:22 +0530
  • 9e189947d1 Minor fix after bd7cc4234d (thanks @belladoreai) oobabooga 2024-05-21 10:37:30 -0700
  • ae86292159 Fix getting Phi-3-small-128k-instruct logits oobabooga 2024-05-21 10:35:00 -0700
  • 21a4ac0845
    Fixed minor typo in docs - Training Tab.md rohitanshu 2024-05-21 22:59:47 +0530
  • bd7cc4234d
    Backend cleanup (#6025) oobabooga 2024-05-21 13:32:02 -0300
  • 5dba2743ff Merge branch 'dev' into backend-cleanup oobabooga 2024-05-21 09:29:14 -0700
  • e2fe90df04 Add chat-templates to .gitignore Touch-Night 2024-05-22 00:06:35 +0800
  • 690ebc533b Save chat template Touch-Night 2024-05-21 23:57:08 +0800
  • de33a036c9 clarify workaround comment marcel 2024-05-21 17:12:26 +0200
  • 9d85533240
    --- updated-dependencies: - dependency-name: optimum dependency-type: direct:production ... dependabot[bot] 2024-05-20 20:25:01 +0000
  • ab1c451efb
    --- updated-dependencies: - dependency-name: gradio dependency-type: direct:production ... dependabot[bot] 2024-05-20 20:24:56 +0000
  • 87125d5833
    --- updated-dependencies: - dependency-name: transformers dependency-type: direct:production ... dependabot[bot] 2024-05-20 20:24:46 +0000
  • b30e830230
    --- updated-dependencies: - dependency-name: hqq dependency-type: direct:production update-type: version-update:semver-patch ... dependabot[bot] 2024-05-20 20:24:41 +0000
  • ae7a644c8f
    --- updated-dependencies: - dependency-name: peft dependency-type: direct:production ... dependabot[bot] 2024-05-20 20:24:31 +0000
  • 6a1682aa95 README: update command-line flags with raw --help output oobabooga 2024-05-19 20:28:21 -0700
  • 852c943769
    DRY: A modern repetition penalty that reliably prevents looping (#5677) Philipp Emanuel Weidmann 2024-05-20 08:23:47 +0530
  • 6177676302 Merge branch 'dev' into p-e-w-dry oobabooga 2024-05-19 19:29:53 -0700
  • 9f77ed1b98
    --idle-timeout flag to unload the model if unused for N minutes (#6026) oobabooga 2024-05-19 23:29:39 -0300
  • 4bc923e29c Organize a bit (put close to no_repeat_ngram_size in the codebase) oobabooga 2024-05-19 19:26:15 -0700
  • 818b4e0354
    Let grammar escape backslashes (#5865) altoiddealer 2024-05-19 19:26:09 -0400
  • 907702c204
    Fix gguf multipart file loading (#5857) Tisjwlf 2024-05-20 01:22:09 +0200
  • d7bd3da35e
    Add Llama 3 instruction template (#5891) Guanghua Lu 2024-05-20 07:17:26 +0800
  • 255dfb2ead Prevent llama-3 derivatives from having wrong template oobabooga 2024-05-19 16:16:29 -0700
  • 5cb59707f3
    fix: grammar not support utf-8 (#5900) A0nameless0man 2024-05-20 07:10:39 +0800
  • 8456d13349
    [docs] small docker changes (#5917) Jari Van Melckebeke 2024-05-20 01:09:37 +0200
  • e10c779eb9 Revert README changes oobabooga 2024-05-19 16:08:25 -0700
  • b63dc4e325
    UI: Warn user if they are trying to load a model from no path (#6006) Samuel Wein 2024-05-20 01:05:17 +0200
  • 2de586f586
    Update accelerate requirement from ==0.27.* to ==0.30.* (#5989) dependabot[bot] 2024-05-19 20:03:18 -0300
  • 6b546a2c8b
    llama.cpp: increase the max threads from 32 to 256 (#5889) chr 2024-05-20 07:02:19 +0800
  • 8519e56aa4
    Bump sse-starlette from 1.6.5 to 2.1.0 dependabot[bot] 2024-05-19 22:02:05 +0000
  • abe5ddc883
    Merge pull request #6027 from oobabooga/dev oobabooga 2024-05-19 19:01:11 -0300
  • d35d92d8f6 Clean up oobabooga 2024-05-19 14:37:02 -0700
  • 2e43780189 --idle-timeout flag to unload the model if unused for N minutes oobabooga 2024-05-19 14:33:26 -0700
  • 13f37b79af
    Merge 49087d67a2 into a38a37b3b3 Nick Potafiy 2024-05-19 16:09:15 -0400
  • a38a37b3b3 llama.cpp: default n_gpu_layers to the maximum value for the model automatically oobabooga 2024-05-19 10:57:42 -0700
  • 0d7e5033f1 Remove obsolete code oobabooga 2024-05-19 09:05:57 -0700
  • 2d2b9308e1 Attempt to fix HQQ oobabooga 2024-05-19 09:00:07 -0700
  • dbb1094b0d Remove GPTQ-for-LLaMa oobabooga 2024-05-19 08:51:32 -0700
  • 89fe454878 Remove inject_fused_attention option from AutoGPTQ to make it functional oobabooga 2024-05-19 07:42:57 -0700
  • a3f8adc9b8 Bump transformers to 4.41 oobabooga 2024-05-19 07:23:28 -0700
  • 569f6b691b Bump AQLM to 1.1.5 oobabooga 2024-05-19 07:22:58 -0700
  • fbc63b39c3 Bump HQQ to 0.1.7.post2 oobabooga 2024-05-19 07:21:42 -0700
  • 0b21656256 Bump AutoGPTQ to 0.7.1, remove ROCm support, remove CUDA 11.8 support oobabooga 2024-05-19 07:20:05 -0700
  • 23928783fc Bump AutoAWQ to 0.2.5 oobabooga 2024-05-19 07:10:30 -0700
  • 6a3f97db61 Remove QuIP# oobabooga 2024-05-19 07:07:28 -0700
  • a4611232b7 Make --verbose output less spammy oobabooga 2024-05-18 09:57:00 -0700
  • 0d90b3a25c Bump llama-cpp-python to 0.2.75 oobabooga 2024-05-18 05:26:26 -0700
  • 7348482ed8
    Create wrangler.toml N3XT-G3N-DEV 2024-05-17 07:45:49 +0200
  • fee7122a58 increase the max threads from 32 to 256 hongruichen 2024-02-20 10:36:02 +0800
  • f9b2ff1616
    Merge branch 'oobabooga:dev' into dev Artificiangel 2024-05-15 06:27:21 -0400
  • 2585d07e2a
    Document DRY parameters in UI Philipp Emanuel Weidmann 2024-05-14 18:19:11 +0530
  • 80cbb8f230
    Support both JSON array notation and comma-separated strings for sequence breakers Philipp Emanuel Weidmann 2024-05-14 17:52:43 +0530
  • fc58c3b0ba
    Bring parameter order in constructor in line with rest of code Philipp Emanuel Weidmann 2024-05-14 17:20:08 +0530
  • 61f0c65eb3
    Update gradio requirement from ==4.26.* to ==4.31.* dependabot[bot] 2024-05-13 20:12:54 +0000
  • 8422027848
    Update api.py Raunak-Kumar7 2024-05-13 12:57:41 +0530
  • d1833271db
    Warn user if they are trying to load a model from no path Samuel Wein 2024-05-12 20:03:30 +0200
  • e225b0b995 downloader: fix downloading 01-ai/Yi-1.5-34B-Chat oobabooga 2024-05-12 10:43:50 -0700
  • 9557f49f2f Bump llama-cpp-python to 0.2.73 oobabooga 2024-05-11 10:53:19 -0700
  • a6420c444c Chat: fix HTML changes observing in main.js Corneliu Zuzu 2024-05-09 16:26:18 +0300