sigoden
72e689fab0
fix: cohere tool use ( #605 )
...
cohere cannot specify both message and tool_results in multistep mode
2024-06-17 09:31:17 +08:00
sigoden
638bf32766
refactor: improve code quatity ( #604 )
2024-06-17 09:14:54 +08:00
sigoden
12872b3d29
refactor: update models.yaml ( #602 )
2024-06-15 06:39:55 +08:00
sigoden
746b087111
refactor: add/modify rag-related config ( #599 )
2024-06-14 19:12:18 +08:00
sigoden
c1d39e4621
chore: clippy and format ( #600 )
2024-06-14 19:06:08 +08:00
sigoden
8cbc74e9ad
refactor: change spinner text for rag searching ( #596 )
2024-06-13 20:05:40 +08:00
sigoden
f9847475b8
feat: add rag and bot related cli options ( #595 )
2024-06-13 19:41:54 +08:00
sigoden
255b194bcc
feat: add .starter repl command ( #594 )
2024-06-13 07:41:29 +08:00
sigoden
64982b4510
feat: rag splitter supports languages ( #593 )
2024-06-12 19:17:40 +08:00
sigoden
492b006db7
refactor: load pdf/docx files ( #592 )
2024-06-12 17:07:49 +08:00
sigoden
fe3a22255d
refactor: 'temp' can not be used as name of saved session ( #591 )
2024-06-12 09:41:05 +08:00
sigoden
b24cb66309
refactor: bot session shouldn't use non-bot last message ( #590 )
2024-06-12 09:21:43 +08:00
sigoden
95221ade46
refactor: bot dir/file path ( #589 )
2024-06-12 08:56:56 +08:00
sigoden
1f33b3a07a
refactor: rag default_chunk_size ( #588 )
2024-06-12 08:31:22 +08:00
sigoden
e9e48f2320
refactor: rename selected_functions to functions_filter ( #587 )
2024-06-12 07:17:57 +08:00
sigoden
c0bfb90b20
refactor: improve openai client ( #586 )
2024-06-11 21:07:32 +08:00
sigoden
aa94fb2f83
fix: infinite loop of function calls on poor LLM ( #585 )
2024-06-11 17:52:04 +08:00
sigoden
b05b730cb5
feat: add config repl_prelude and bot_prelude ( #584 )
2024-06-11 15:44:35 +08:00
sigoden
fcdfeea548
refactor: rename function_matcher to selected_functions ( #583 )
2024-06-11 14:28:33 +08:00
sigoden
8615293747
feat: add config dangerously_functions ( #582 )
2024-06-11 14:01:45 +08:00
sigoden
822688a06a
refactor: adjust config methods order ( #581 )
2024-06-11 11:46:29 +08:00
sigoden
bb867c4fcb
feat: support bot ( #579 )
...
* feat: support bots
* refactor with RoleLike
* improve exiting session
* make bot works with rag
* refactor repl assert state
* add bot banner
* repl complete bots according bots.txt
* fix on windows
* remove threadpool executing function callings
* adjust repl left_prompt
* move bot config to global config.yaml
* `.bot` throw err if funciton callings is not configured
2024-06-11 11:00:12 +08:00
sigoden
5635ca6a58
refactor: warning model or client don't support function calling ( #572 )
2024-06-06 06:23:54 +08:00
sigoden
93ef6da629
refactor: global IS_STDOUT_TERMINAL ( #571 )
2024-06-06 06:08:41 +08:00
sigoden
c26594c3c8
refactor: config refinement ( #570 )
2024-06-06 05:27:21 +08:00
sigoden
171a46eee2
refactor: change error message of the model does not support vision ( #568 )
2024-06-05 21:21:20 +08:00
sigoden
ce8d6d750d
revert: feat: use systemInstruction for gemini-1.5-* models ( #564 ) ( #567 )
2024-06-05 21:16:06 +08:00
sigoden
4b27378443
feat: custom more path to file/dirs with environment variables ( #565 )
2024-06-05 15:18:34 +08:00
sigoden
d4e043e5bf
feat: use systemInstruction for gemini-1.5-* models ( #564 )
2024-06-05 15:07:21 +08:00
sigoden
b17719457b
chore: format
2024-06-05 06:06:11 +00:00
sigoden
eacc88f04a
chore: update readme
2024-06-05 04:37:08 +00:00
sigoden
a732291f33
refactor: rename pass_max_tokens to require_max_tokens ( #562 )
2024-06-05 09:10:12 +08:00
sigoden
1ec6abfaee
feat: support RAG ( #560 )
...
* feat: support RAG
* support more embeddings models and implement concurrent embedding api
* show the progress of addings paths
* ignore embedding context when saving message
* embedding model max_chunk_size => default_chunk_size
* support pdf and pandoc formats (docx, epub, ipynb)
2024-06-05 09:02:23 +08:00
sigoden
71f2e94579
refactor: switch to bitflags State ( #557 )
2024-06-02 19:27:41 +08:00
sigoden
571d1022f6
refactor: rename some client structs and methods ( #555 )
...
* rename `Completeion*` to `ChatCompletions*`
* rename `send_message*` to `chat_completions*`
* rename `request_builder` to `chat_completions_builder`
* rename `build_body` to `build_chat_completions_body`
* rename `extract_completion` to `extract_chat_completions`
* format
* remove unused config fields
2024-06-01 17:47:49 +08:00
sigoden
259583f4f7
chore: update readme
2024-06-01 03:30:34 +00:00
sigoden
38797e3264
chore: release v0.18.0 ( #551 )
2024-06-01 10:45:53 +08:00
sigoden
54a837784c
refactor: rename SendData to CompletionData ( #553 )
2024-05-30 17:10:02 +08:00
sigoden
fa4bf14e02
refactor: adjust default compress_threshold to 4000 ( #552 )
2024-05-30 10:19:29 +08:00
sigoden
c0f4d2d712
chore: fix typos
2024-05-30 01:02:11 +00:00
ProjectMoon
00f3cb182f
refactor: use json_stream for ollama to improve reliability ( #549 )
...
* Use JSON stream for ollama to improve reliability. Fixes #548 .
* remove unused import
* fix clippy error
* format
---------
Co-authored-by: sigoden <sigoden@gmail.com>
2024-05-29 20:27:07 +08:00
sigoden
4fa92b020a
refactor: temp session ignore save_session ( #547 )
2024-05-29 07:06:05 +08:00
rolfwilms
569317728c
fix: bedrock issues ( #544 )
...
* Removed extraneous key [stream] for AWS Bedrock Claude models.
* Reduceddefault AWS Bedrock llama-3 max_output_tokens to 2048 to align with API requirements.
---------
Co-authored-by: Rolf Wilms <rwilms@csc.com>
2024-05-26 05:52:24 +08:00
sigoden
5458150ed3
fix: json stream parser and refine client modules ( #538 )
2024-05-23 19:28:56 +08:00
sigoden
2ccbb0f06a
refactor: qiawen client add qwen-long ( #537 )
2024-05-23 12:26:24 +08:00
sigoden
50b13d2de9
refactor: rename function_filter to function_matcher ( #535 )
2024-05-22 21:37:24 +08:00
sigoden
ba3bcfd67c
feat: allow patching req body with client config ( #534 )
2024-05-22 21:29:23 +08:00
sigoden
91a06543b2
refactor: adjust some placeholder/prompt text ( #530 )
2024-05-21 08:06:57 +08:00
sigoden
2174dc055f
refactor: change prompt for executing tool functions ( #529 )
...
* refactor: change prompt for executing tool functions
* add AICHAT_FUNCTION_AUTO_EXECUTE to auto confirm
2024-05-21 07:56:02 +08:00
sigoden
d4de2d761b
refactor: no export LLM_FUNCTION_DATA, pass json as command arg ( #528 )
2024-05-20 17:27:34 +08:00
