Commit Graph

12 Commits

Author SHA1 Message Date
extio-de
1c94b97b92 DatasetTool marker interface; 2024-12-18 20:42:12 +01:00
extio-de
75c2109d6d Switch dataset preparer to lmlib; 2024-12-18 20:05:03 +01:00
extio-de
f0872fb7ee createCharacterNameMapping: Changed log level; 2024-12-02 11:36:03 +01:00
extio-de
57f7ec2e48 Text2Jsonl2WithContextualPrompts: Prompt improved; Name mapping requires at least 5 characters (otherwise we may replace random syllables); 2024-12-01 22:50:53 +01:00
extio-de
b027c2a892 Text2Jsonl2WithContextualPrompts: Rename persons; Prompt conversation contexts; 2024-12-01 18:32:36 +01:00
extio-de
dda9400cc1 Text2Jsonl2WithContextualPrompts: Overlap chunks; 2024-11-16 10:54:06 +01:00
extio-de
82f969d666 Fix paragraphToContextualPrompt response post-processing; 2024-11-03 22:53:35 +01:00
extio-de
cb8ac57580 Improved prompt for Text2Jsonl2WithContextualPrompts; 2024-11-03 20:37:22 +01:00
extio-de
509f36238d Fix MixtralPromptStrategy; Refactoring; 2024-11-02 12:24:28 +01:00
extio-de
65e50e1dcc Sliding contexts for contextual prompts; 2024-11-01 15:08:26 +01:00
extio-de
7232387134 Switched remaining tools to streaming output;
Generate contextual prompt datasets as single files per input;
Added Alpaca prompt template;
Generate multiple contextual prompt lengths;
2024-11-01 13:43:59 +01:00
extio-de
efb520fc07 New generator: Jsonl QA style contextual prompts;
JsonL: Add sliding windows;
Streamed jsonl output;
Memory optimization: Buffered, blocking task creation;
2024-10-29 23:51:41 +01:00