Commit Graph

  • 4883e20fa7
    Fix openai extension script.py - TypeError: '_Environ' object is not callable (#1753) Thireus ☠ 2023-05-03 13:51:49 +0100
  • f54256e348 Rename no_mmap to no-mmap oobabooga 2023-05-03 09:50:31 -0300
  • 4dcd569723
    Add a load model button LaaZa 2023-05-03 10:08:40 +0000
  • f2bffedd4e
    Update README.md Priestru 2023-05-03 14:06:36 +0500
  • 9f923bdd11
    Update api-example-stream.py EvolventaAGG 2023-05-03 13:51:07 +0500
  • 88df783471
    Epsilon, eta for api EvolventaAGG 2023-05-03 13:49:48 +0500
  • eb91dd966e
    Merge branch 'oobabooga:main' into main EvolventaAGG 2023-05-03 13:43:33 +0500
  • 78b83294fe
    Epsilon, eta for api EvolventaAGG 2023-05-03 13:40:56 +0500
  • 127bcaf249
    Openai Chat Completion support improvement Thireus ☠ 2023-05-03 08:57:31 +0100
  • 57699beb42
    Merge branch 'oobabooga:main' into AutoGPTQ LaaZa 2023-05-03 07:16:02 +0000
  • 17dd7e375b add edits endpoint Matthew Ashton 2023-05-03 03:10:33 -0400
  • c418e0b254
    Fix openai extension script.py - TypeError: '_Environ' object is not callable Thireus ☠ 2023-05-03 07:05:52 +0100
  • dec31af910
    Create .gitignore (#43) Roberts Slisans 2023-05-03 05:47:19 +0300
  • 24c5ba2b9c
    Fixed error when $OS_ARCH returns aarch64 (#45) Semih Aslan 2023-05-03 02:47:03 +0000
  • 875da16b7b Minor CSS improvements in chat mode oobabooga 2023-05-02 23:38:51 -0300
  • e3968f7dd0
    Fix Training Pad Token (#1678) practicaldreamer 2023-05-02 21:16:08 -0500
  • 80c2f25131
    LLaVA: small fixes (#1664) Wojtab 2023-05-03 04:12:22 +0200
  • c31b0f15a7 Remove some spaces oobabooga 2023-05-02 23:07:07 -0300
  • 320fcfde4e Style/pep8 improvements oobabooga 2023-05-02 23:05:38 -0300
  • ecd79caa68
    Update Extensions.md oobabooga 2023-05-02 22:52:32 -0300
  • 7ac41b87df
    add openai compatible api (#1475) matatonic 2023-05-02 21:49:53 -0400
  • f5651bcb9e
    Merge branch 'oobabooga:main' into AutoGPTQ LaaZa 2023-05-02 23:10:50 +0000
  • 4e09df4034 Only show extension in UI if it has an ui() function oobabooga 2023-05-02 19:20:02 -0300
  • d016c38640 Bump gradio version oobabooga 2023-05-02 19:19:33 -0300
  • 88cdf6ed3d Prevent websocket from disconnecting oobabooga 2023-05-02 19:03:19 -0300
  • fbcd32988e
    added no_mmap & mlock parameters to llama.cpp and removed llamacpp_model_alternative (#1649) Ahmed Said 2023-05-03 00:25:28 +0300
  • 67e99c7d12 Add a space oobabooga 2023-05-02 18:25:07 -0300
  • 2974f55df0 Rename the file oobabooga 2023-05-02 18:23:25 -0300
  • 7fc6fdf753 Clean up oobabooga 2023-05-02 18:21:44 -0300
  • a83e665572 fix save_audio (#1648) subversion23 2023-05-02 22:14:58 +0200
  • 318538e862 simplify embeddings model download Matthew Ashton 2023-05-02 14:31:01 -0400
  • 00009a61d5 Merge branch 'main' into correct-llava-projector Wojtab 2023-05-02 18:34:12 +0200
  • 8df5861dad Merge branch 'main' into generalize-multimodality Wojtab 2023-05-02 17:39:01 +0200
  • 4babb22f84
    Fix/Improve a bunch of things (#42) Blake Wyatt 2023-05-02 11:28:20 -0400
  • ba4f082389 Add support for Triton LLaMA speed optimisations. And re-enable cpu support for testing. LaaZa 2023-05-02 18:07:21 +0300
  • 125c1bee54 Automatically detect wbits and groupsize CarlKenner 2023-05-02 22:04:08 +0930
  • 728a207433
    Merge branch 'oobabooga:main' into AutoGPTQ LaaZa 2023-05-02 09:51:59 +0000
  • a859b40859 minor embeddings update, chatgpt-api added Matthew Ashton 2023-05-02 05:18:28 -0400
  • 3c60b50489
    Merge branch 'oobabooga:main' into chat-api Xan 2023-05-02 16:39:00 +1000
  • 11d9706ffa
    Merge branch 'oobabooga:main' into main matatonic 2023-05-02 00:49:47 -0400
  • 2f1a2846d1
    Verbose should always print special tokens in input (#1707) Carl Kenner 2023-05-02 13:54:56 +0930
  • 0df0b2d0f9
    optimize stopping strings processing (#1625) Alex "mcmonkey" Goodwin 2023-05-01 21:21:54 -0700
  • f653530243 spaces ocd oobabooga 2023-05-02 01:18:03 -0300
  • 58474d26dd Add file for comparison oobabooga 2023-05-02 01:06:15 -0300
  • e6a78c00f2
    Update Docker.md oobabooga 2023-05-02 00:51:10 -0300
  • 3c67fc0362
    Allow groupsize 1024, needed for larger models eg 30B to lower VRAM usage (#1660) Tom Jobbins 2023-05-02 04:46:26 +0100
  • 78bd4d3a5c
    Update LLaMA-model.md (#1700) Lawrence M Stewart 2023-05-01 20:44:09 -0700
  • f659415170
    fixed variable name "context" to "prompt" (#1716) Dhaladom 2023-05-02 05:43:40 +0200
  • 280c2f285f
    Bump safetensors from 0.3.0 to 0.3.1 (#1720) dependabot[bot] 2023-05-02 00:42:39 -0300
  • 56b13d5d48 Bump llama-cpp-python version oobabooga 2023-05-02 00:41:54 -0300
  • f1ac79da8c Bump llama-cpp-python version oobabooga 2023-05-02 00:40:59 -0300
  • 9cc054ff30 fix Wojtab 2023-05-02 05:37:01 +0200
  • 75006a9536 add multimodal Wojtab 2023-05-02 05:33:28 +0200
  • dbe4930709
    Bump safetensors from 0.3.0 to 0.3.1 dependabot[bot] 2023-05-01 21:07:15 +0000
  • 41945f0d2b
    Bump llama-cpp-python from 0.1.36 to 0.1.40 dependabot[bot] 2023-05-01 21:07:04 +0000
  • 8705f64f50
    Bump gradio from 3.25.0 to 3.28.1 dependabot[bot] 2023-05-01 21:06:57 +0000
  • 3ea78769eb Add a flag (--dont-load-cpu-model) to allow a user to explicitly skip using CPU based models, if they're present in the model directory. phalanx-anon 2023-05-01 14:41:40 -0600
  • ba838b734b
    fixed variable name "context" to "prompt" Dhaladom 2023-05-01 22:17:23 +0200
  • 82b84f959c Update requirements.txt catalpaaa 2023-05-01 11:45:23 -0700
  • 4f156e4e54 More robust quantize config parsing and setting for act-order or desc_act LaaZa 2023-05-01 21:06:25 +0300
  • 6eb686cb1e adds train_dataset_sample log to training.py practicaldreamer 2023-05-01 11:26:36 -0500
  • 9db05f611c Verbose should always print special tokens in input CarlKenner 2023-05-01 22:31:17 +0930
  • 212e2fba65 debug off Matthew Ashton 2023-05-01 03:54:57 -0400
  • 53c4244bbc legacy generate, batch embeddings + better model Matthew Ashton 2023-05-01 03:54:22 -0400
  • 02d00467aa Added recommended changes minipasila 2023-05-01 06:34:27 +0300
  • b173274c63 Check for quantize_config.json and set wbits and groupsize according to it. LaaZa 2023-05-01 06:15:43 +0300
  • 96cea8490d
    Update LLaMA-model.md Lawrence M Stewart 2023-04-30 19:43:55 -0700
  • e9d2697f51 remove custom_generate_chat_prompt from LLaVA Wojtab 2023-05-01 02:57:44 +0200
  • 4b542ec2f1 LLaVA 7B just dropped, add to readme that there is no support for it currently Wojtab 2023-04-30 22:04:30 +0200
  • 0fd4857646 Implement offloading and splitting between multiple devices. LaaZa 2023-04-30 21:32:31 +0300
  • 0246f8842b compatibility, legacy, maybe babyagi, openai-node Matthew Ashton 2023-04-30 12:49:29 -0400
  • 109a83044a
    added Metharme minipasila 2023-04-30 18:55:42 +0300
  • 753a29bc4b
    changed things minipasila 2023-04-30 18:52:06 +0300
  • d4b7a936b6
    Update script.py ezra-ch 2023-04-30 21:25:10 +0800
  • e9ae49c135
    Rename Metharme to Metharme.yaml minipasila 2023-04-30 15:33:28 +0300
  • eeb7ce1136
    Fix Training Pad Token practicaldreamer 2023-04-30 03:56:18 -0500
  • 494df29109
    Added Metharme minipasila 2023-04-30 11:55:48 +0300
  • 016eb0c43b some langchain improvement, clamp some inputs Matthew Ashton 2023-04-30 04:29:19 -0400
  • d17bf85478
    Fix Training Pad Token practicaldreamer 2023-04-30 02:14:29 -0500
  • ea22a3efb5 moderations v0, legacy support, discord bot, perf Matthew Ashton 2023-04-30 02:20:46 -0400
  • bb1b1b128c Disable cpu support for now. Something in the inference path assumes gpu. LaaZa 2023-04-30 05:35:46 +0300
  • 38ab214a41 add API example Wojtab 2023-04-30 03:51:05 +0200
  • 8be652db63 fix stopping strings if tokenizer extension adds/removes tokens Wojtab 2023-04-30 03:28:56 +0200
  • eb4f7f49df Implement support for AutoGPTQ for loading GPTQ quantized models. LaaZa 2023-04-30 04:13:38 +0300
  • 0f85a4bc50 remove reference to custom stopping strings from readme Wojtab 2023-04-30 02:20:05 +0200
  • ba439c09d4 change multimodal projector to the correct one Wojtab 2023-04-30 02:01:05 +0200
  • b75659c54e Allow downloading model from HF branch via UI TheBloke 2023-04-30 00:10:01 +0100
  • 8bb0c7874a Allow groupsize 1024, needed for larger models eg 30B to lower VRAM usage TheBloke 2023-04-29 22:43:33 +0100
  • dfe550852e tested all-mpnet-base-v2 model Matthew Ashton 2023-04-29 16:34:00 -0400
  • 4b03d43233 embeddings w/ SentenceTransformer/all-MiniLM-L6-v2 Matthew Ashton 2023-04-29 15:54:12 -0400
  • 6a2bc13227 Remove mutable defaults from function signature. IJumpAround 2023-04-29 15:48:31 -0400
  • 66e4454c19 update README and modify no_mmmap handling. Ahmed Said 2023-04-29 20:41:10 +0200
  • 7d89891af7 Move Stream to the first of the stop_criteria list, to make stop_string appear in output completely SilverJim 2023-04-30 00:43:41 +0800
  • afe955392d Fix stopping_string issue by Creating RegxStoppingCriteria and use RegxStoppingCriteria in stopping_string SilverJim 2023-04-30 00:03:41 +0800
  • 435e947c62
    Update llama-cpp-python whl to 0.1.39 Thireus ☠ 2023-04-29 13:12:43 +0100
  • 1875f78dfa
    Update llama-cpp-python to 0.1.39 for q5 support Thireus ☠ 2023-04-29 13:08:34 +0100
  • d056280664 added no_mmap & mlock parameters to llama.cpp Ahmed Said 2023-04-29 12:48:02 +0200
  • 8cb602a8a8 Implement basic chat API and api-example Xan 2023-04-29 16:04:37 +1000
  • 64fd5c70d3
    Update README.md bgrmwbys 2023-04-28 16:37:52 -0400
  • c29fdc3461 Update requirements.txt catalpaaa 2023-04-28 12:17:27 -0700