0.3.0

0.3.0

• Added textmodel_gepa_classify and a predict and summary function for it
• import localllm_connect as connect making sure you can do
  import localllm
  localllm.connect("localllm/modelxyz")
• Move converters to localllm.utilities.converters
• model_name in LocalLLM objects use now the name passed on to localllm_download_model, not the gguf filename
• localllm_download_model, added:
  • LFM2.5-8B-A1B-Q4_0
  • gemma-4-E4B-it-qat-UD-Q4_K_XL
  • gemma-4-12B-it-Q4_K_M
  • gemma-4-12B-it-qat-UD-Q4_K_XL

0.2.2

0.2.2

• Added optional converters
  • from localllm.converters import tif, convert_dspy_example
  • added dependency on s3generics
  • added optional dependency on pandas for the converters
• Bump dependency to Python >=3.10
• LocalChatAdapter: more lenient parsing of the output fields [## value# ] and variants
• localllm_download_model, added:
  • gemma-4-E4B-it-Q8_0
  • gemma-4-E4B-it-Q4_K_M
  • translategemma-12b-it-Q4_K_M
  • gemma-3-4b-it-qat-Q4_0
• localllm_download_model, renamed
  • gemma-4-e2b-it-Q8_0 to gemma-4-E2B-it-Q8_0
• localllm_download_model - reorder by bigger to smaller
• Add localllm_list_models to list up the models which you can download easily

0.2.1

0.2.1

• Exported localllm_connect
• localllm_download_model, added:
  • Qwen3.5-9B-Q4_K_M
  • Qwen3.5-4B-Q4_K_M
  • Qwen3.5-2B-Q4_K_M
  • Qwen3.5-0.8B-Q8_0
  • LFM2.5-350M-Q8_0
  • gemma-4-e2b-it-Q8_0
  • gemma-4-E2B-it-Q4_K_M
• LocalLLM
  • put call in try-catch block and return [None] in case of an error (e.g. context to big for the model) + log the error in case of trace
  • in call trace more arguments if requested

0.2.0

0.2.0

• Added LocalChatAdapter (extending BAMLAdapator) to parse out the LLM response which are wrongly formatted more gracefully defaulting to dspy_signature_defaults in case of complete failure
• Added dspy_signature_defaults
• localllm_download_model, added:
  • gemma-3-12b-it-qat-Q4_0
  • Qwen3-8B-Q4_K_M
  • Qwen3-4B-Instruct-Q4_K_M
  • LFM2.5-1.2B-Instruct-Q4_K_M
  • LFM2.5-1.2B-Instruct-Q8_0
• Added localllm.config.localllm_connect
• Bump dependencies to llama-cpp-python>=0.3.8 and dspy>=3.0.0

0.1.0

0.1.0

• Added functionality to download a model: localllm_download_model
• Added functionality run a local model (with llama-cpp) alongside dspy: LocalLLM
• localllm_download_model:
  • gemma-3-270m-it-Q8_0
  • gemma-3-270m-it-qat-Q4_0
  • gemma-3-1b-it-Q8_0
  • gemma-3-4b-it-Q4_K_M
  • GLM-4.6V-Flash-Q4_K_M
  • translategemma-4b-it-q8_0