v0.9.0

0.9.0 (2025-02-26)

new:

• Support for using self-ask prompt strategy with RAG (#120)
• Improved table understanding when invoking LLm.ask. (#124)
• helpers for document metadata (#121)

changed

• Added k and score_threshold arguments to LLM.ask (#122)
• Added n_proc parameter to control the number of CPUs used
  by LLM.ingest (ee09807)
• Upgrade version of chromadb (#125)

fixed:

• Ensure table-processing is sequential and not parallelized (#123)
• Fixes to support newer version of langchain_community. (#125)

v0.8.0

0.8.0 (2025-02-13)

new:

• Added HFClassifier to pipelines.classifier module (#119)
• Added SKClassifier to pipelines.classifier module (#118)
• sk "helper" module to fit simple scikit-learn text models (#117)

changed

• Added process_documents function (#117)

fixed:

• Pass autodetect_encoding argument to TextLoader (#116)

0.7.1

0.7.1 (2024-12-18)

new:

• N/A

changed

• N/A

fixed:

• Fix for HF chat template issue resulting in bad prompt format and bad output for HF transformer models (#113/#114)

0.7.0

0.7.0 (2024-12-16)

new:

• Support for structured outputs (#110)
• Support for table extraction (#106, #107)
• Facilitate identifying tables extracted as HTML (#112)

changed

• Remove dependnency on deprecated RetrievalQA (#108)
• Refactored code base (#109)
• Use new JSON-safe formatting of prompt templates (#109)

fixed:

• Added utils.format_string function to help format template strings
  with embedded JSON (#105)
• support stop strings with transformers (#111)

0.6.1

0.6.1 (2024-12-04)

new:

• N/A

changed

• Changed pdf_use_unstructured to pdf_unstructured and
  pdf2md to pdf_markdown (#102)

fixed:

• N/A

0.6.0

0.6.0 (2024-12-3)

new:

• Improved PDF text extraction including optional markdown
  conversion, table inference, and OCR (#100)

changed

• N/A

fixed:

• Add support for HF training (#98)
• Default to localhost in Web app (#99)

0.5.2

0.5.2 (2024-011-25)

new:

• N/A

changed

• N/A

fixed:

• Allow all Hugging Face pipeline/model arguments to be supplied (#96)

0.5.1

0.5.1 (2024-11-22)

new:

• N/A

changed

• Refactored Hugging Face transformers backend (#95)

fixed:

• Suppress swig deprecation warning (#93)
• Raise error if summarizers encounter bad document (#94)

0.5.0

0.5.0 (2024-11-20)

new:

• Support for Hugging Face transformers as LLM engine instead of Llama.cpp

changed

• LLM.prompt now accepts OpenAI-style messages in form of list of dictionaries

fixed:

• Remove unused imports (#92)

0.4.0

0.4.0 (2024-11-13)

new:

• Added default_model parameter to LLM to more easily use Llama-3.1-8B-Instruct.

changed

• N/A

fixed:

• N/A