Commit Graph

  • 643b53e347 Create FUNDING.yml Brian Dashore 2024-05-09 19:00:41 +00:00
  • c4f7af160e Merge pull request #101 from Bakharovsky/fix_exllamav2_cuda_version Brian Dashore 2024-05-08 16:32:22 -04:00
  • 33c86be45c Update pyproject.toml Arseniy Bakharovsky 2024-05-08 03:31:15 +04:00
  • ae879a623f Main: Add await to an async function kingbri 2024-05-02 21:24:43 -04:00
  • ab526f7278 Revert "API: Remove unncessary Optional signatures" kingbri 2024-05-02 21:23:48 -04:00
  • 7556dcf134 API: Remove unncessary Optional signatures kingbri 2024-04-30 23:51:28 -04:00
  • ae75db1829 Downloader: Cleanup on exception kingbri 2024-04-30 23:26:22 -04:00
  • e4084b15c1 Downloader: Format kingbri 2024-04-30 01:16:41 -04:00
  • 50e0b71690 Downloader: Fix handling of include pattern kingbri 2024-04-30 01:13:06 -04:00
  • 21a01741c9 Downloader: Add include and exclude parameters kingbri 2024-04-30 00:58:54 -04:00
  • c47869c606 Downloader: Fix fallback mechanisms kingbri 2024-04-29 23:33:37 -04:00
  • 55ccd1baad API: Add HuggingFace downloader kingbri 2024-04-29 01:12:30 -04:00
  • 6114bfd221 API: Fix banned_tokens string when empty kingbri 2024-04-28 12:46:28 -04:00
  • 72dff0b6d5 Update README kingbri 2024-04-28 11:26:20 -04:00
  • fb01b164d8 Dependencies: Update flash attention 2 kingbri 2024-04-28 10:52:58 -04:00
  • 0e015ad58e Dependencies: Update ExllamaV2 kingbri 2024-04-28 00:44:03 -04:00
  • 3de93d7c0a Dependencies: Update torch kingbri 2024-04-28 00:43:25 -04:00
  • 4daa6390a5 Dependencies: Unpin lm-format-enforcer kingbri 2024-04-28 00:41:19 -04:00
  • 6f9da97114 API: Add banned_tokens kingbri 2024-04-28 00:40:34 -04:00
  • 5750826120 Model: Remove extraneous print kingbri 2024-04-25 18:49:09 -04:00
  • fb1d2f34c1 OAI: Add response_prefix and fix BOS token issues in chat completions kingbri 2024-04-25 00:54:43 -04:00
  • ed7cd3cb59 Network: Fix socket check timeout kingbri 2024-04-22 21:33:41 -04:00
  • 1e56d43772 Dependencies: Update lm-format-enforcer kingbri 2024-04-22 21:33:28 -04:00
  • 88b0b6f4f1 Model: Cast autosplit_reserve to int kingbri 2024-04-21 23:47:37 -04:00
  • cab789e685 Templates: Migrate to class kingbri 2024-04-21 23:28:14 -04:00
  • 9f93505bc1 OAI: Add skip_special_tokens parameter kingbri 2024-04-21 00:37:46 -04:00
  • 67f061859d Tree: Add transformers_utils kingbri 2024-04-20 00:07:39 -04:00
  • 8824ea0205 Model: Add EOS token support from generation_config.json kingbri 2024-04-19 22:52:32 -04:00
  • 933c5afef0 Dependencies: Update ExllamaV2 and lm-format-enforcer kingbri 2024-04-19 21:15:50 -04:00
  • 65871ebc0c Docker: Add var to pull on build kingbri 2024-04-19 21:06:34 -04:00
  • 209f0370b4 Docker: Switch image and copy config kingbri 2024-04-15 18:01:56 -04:00
  • a2a2e4b866 Merge pull request #94 from pabl-o-ce/docker Brian Dashore 2024-04-15 18:01:09 -04:00
  • 515b3c2930 OAI: Tokenize chat completion messages kingbri 2024-04-15 14:17:16 -04:00
  • ed05f376d9 Dependencies: Switch to LM-format-enforcer fork kingbri 2024-04-14 11:59:49 -04:00
  • 3d14283fe0 Start: Lint kingbri 2024-04-13 12:25:41 -04:00
  • 4d158dac90 Start: Fix when reading from gpu_lib file kingbri 2024-04-13 12:24:30 -04:00
  • 2a0aaa2e8a OAI: Add ability to pass extra vars in jinja templates kingbri 2024-04-11 00:55:32 -04:00
  • b1f3baad74 OAI: Add response_format parameter kingbri 2024-04-09 21:23:01 -04:00
  • de41e9f7e9 Start: Add gpu_lib argument kingbri 2024-04-08 23:28:30 -04:00
  • d759a15559 Model: Fix chunk size handling kingbri 2024-04-07 18:10:50 -04:00
  • 30c4554572 Requirements: Update Exllamav2 kingbri 2024-04-07 18:00:56 -04:00
  • 46ac3beea9 Templates: Support list style chat_template keys kingbri 2024-04-07 11:17:55 -04:00
  • 5bb4995a7c API: Move OAI to APIRouter kingbri 2024-04-06 01:24:54 -04:00
  • 8bdc19124f Start: Fix gpu lib when reading from file kingbri 2024-04-02 22:03:08 -04:00
  • cdb96e4f74 Merge pull request #93 from AlpinDale/chore/log-level Brian Dashore 2024-04-02 00:52:06 -04:00
  • f9f8c97c6d Templates: Fix stop_string parsing kingbri 2024-04-02 00:44:04 -04:00
  • 8a5a82baec Update Dockerfile PΔBLØ ᄃΞ 2024-04-01 22:27:11 -05:00
  • 85271e2b7d fix: Dockerfile work on pyproject.toml PΔBLØ ᄃΞ 2024-04-01 19:32:42 -05:00
  • 1650e6e640 ruff AlpinDale 2024-04-01 23:11:30 +00:00
  • 5e599ddbd4 typo AlpinDale 2024-04-01 23:08:28 +00:00
  • 6c4a1a9c70 make log level a global var AlpinDale 2024-04-01 23:07:30 +00:00
  • 031349133b properly order imports AlpinDale 2024-04-01 23:03:16 +00:00
  • e90ead3b35 chore: make log level configurable via env variable AlpinDale 2024-04-01 22:57:56 +00:00
  • 6ecce1604b Model: Fix log if exl2 version is too low kingbri 2024-03-31 23:11:21 -04:00
  • f534930270 Dependencies: Bump Exllamav2 kingbri 2024-03-31 23:10:28 -04:00
  • d716527b92 Sampling: Add additive param to overrides kingbri 2024-03-31 01:09:59 -04:00
  • 05b5700334 Dependencies: Update torch kingbri 2024-03-30 17:03:37 -04:00
  • 5c94894a1a Dependencies: Update Flash Attention kingbri 2024-03-30 16:58:24 -04:00
  • b11aac51e2 Model: Add torch.inference_mode() to generator function kingbri 2024-03-30 10:42:43 -04:00
  • e8b6a02aa8 API: Move prompt template construction to utils kingbri 2024-03-29 02:24:13 -04:00
  • 190a0b26c3 Model: Fix generation when stream = false kingbri 2024-03-29 02:15:56 -04:00
  • d4280e1378 Dependencies: Add pytorch-triton-rocm kingbri 2024-03-28 11:02:56 -04:00
  • 271f5ba7a4 Templates: Modify alpaca and chatml kingbri 2024-03-27 22:28:41 -04:00
  • dc456f4cc2 Templates: Add stop_strings meta param kingbri 2024-03-27 22:07:43 -04:00
  • 277c540c98 Colab: Update kingbri 2024-03-24 21:48:48 -04:00
  • db62d1e649 OAI: Log request errors to console kingbri 2024-03-23 20:29:17 -04:00
  • 26496c4db2 Dependencies: Require tokenizers kingbri 2024-03-23 01:12:21 -04:00
  • 1755f284cf Model: Prompt users to install extras if dependencies don't exist kingbri 2024-03-22 22:13:55 -04:00
  • f952b81ccf API: Remove uvicorn signal handler injection kingbri 2024-03-21 23:23:45 -04:00
  • 6dfcbbd813 Common: Migrate request utils to networking kingbri 2024-03-21 23:21:57 -04:00
  • 2961c5f3f9 API: Handle request disconnect on non-streaming gens kingbri 2024-03-21 23:12:59 -04:00
  • 44b7319710 Start: Print pip install command kingbri 2024-03-21 18:13:41 -04:00
  • 5055a98e41 Model: Wrap load in inference_mode kingbri 2024-03-21 18:04:22 -04:00
  • 37a80334a8 Dependencies: Add packaging kingbri 2024-03-21 11:27:27 -04:00
  • 56fdfb5f8e OAI: Add stream to gen params kingbri 2024-03-21 00:55:44 -04:00
  • 69e41e994c Model: Fix generation with non-streaming and logprobs kingbri 2024-03-21 00:47:24 -04:00
  • 345bcc30c7 Dependencies: Add extras feature kingbri 2024-03-21 00:02:13 -04:00
  • 51b289cab2 Actions: Fix workflows kingbri 2024-03-20 15:15:34 -04:00
  • 1e7cf1e5a4 Start: Prompt user for GPU/lib kingbri 2024-03-20 14:05:04 -04:00
  • 7e669527ed Model: Fix tokenizer bugs kingbri 2024-03-20 11:23:17 -04:00
  • 07d9b7cf7b Model: Add abort on generation kingbri 2024-03-20 11:01:39 -04:00
  • 7020a0a2d1 Dependencies: Update Exllamav2 kingbri 2024-03-20 11:01:09 -04:00
  • 061e1d94c2 Ruff: Migrate to pyproject kingbri 2024-03-20 00:55:31 -04:00
  • 1059101b23 Dependencies: Remove requirements-*.txt files kingbri 2024-03-20 00:47:53 -04:00
  • 72b08624a3 Start: Update to use pyproject kingbri 2024-03-20 00:46:30 -04:00
  • b1ca435695 Tree: Add pyproject.toml kingbri 2024-03-19 23:48:34 -04:00
  • b74603db59 Model: Log metrics before yielding a stop kingbri 2024-03-20 01:17:04 -04:00
  • 09a4c79847 Model: Auto-scale max_tokens by default kingbri 2024-03-18 22:54:59 -04:00
  • 8cbb59d6e1 Model: Cleanup some comments kingbri 2024-03-18 22:20:45 -04:00
  • 4f75fb5588 Model: Adjust max output len kingbri 2024-03-18 22:10:00 -04:00
  • 2704ff8344 Tree: Format kingbri 2024-03-18 16:02:29 -04:00
  • 5c7fc69ded API: Fix finish_reason returns kingbri 2024-03-18 15:59:28 -04:00
  • 25f5d4a690 API: Cleanup permission endpoint kingbri 2024-03-18 15:13:26 -04:00
  • 3c08f46c51 Endpoints: Add key permission checker kingbri 2024-03-18 00:45:40 -04:00
  • c9a6d9ae1f Model: Switch to begin_stream_ex kingbri 2024-03-17 14:41:16 -04:00
  • 08bcc6307a Config: Update description part 2 kingbri 2024-03-17 01:07:23 -04:00
  • 7abbac098a Config: Update Q4 in comments kingbri 2024-03-17 01:04:12 -04:00
  • 14d8ec2007 Signal: Fix signal handlers for uvicorn kingbri 2024-03-16 23:18:27 -04:00
  • 95e44c20d6 Model: Fix load if model didn't load properly kingbri 2024-03-16 22:57:45 -04:00
  • 2755fd1af0 API: Fix blocking iterator execution kingbri 2024-03-16 22:31:50 -04:00