Agus
Agus
**Is your feature request related to a problem? Please describe.** The `LLM` classes could use the docstrings from the parent class, see for example `AnyscaleLLM`. **Describe the solution you'd like**...
**Is your feature request related to a problem? Please describe.** In #601 we added the possibility to generate structured outputs from `llama-cpp`, `transformers` and `vllm` using `outlines`. It would be...
**Is your feature request related to a problem? Please describe.** The implementation of the structured outputs in #601 only allows for a single structure (json schema, `BaseModel` in case of...
## Description This PR implements cache at step level. Previously, we computed a signature for a pipeline, and when this signature changed, we recomputed everything. Now the idea is to...
## Description ⚠️ Work in progress This PR improves the `FormatTextGenerationSFT` task to allow preparing fine tuning datasets with function calling.
[FEATURE] Allow `FormatTextGenerationSFT` to include tools/function calls in the formatted messages.
**Is your feature request related to a problem? Please describe.** With the new `StructuredGeneration` task we can generate datasets for function calling, and we could simplify preparing the dataset for...
## Description When generating a name for the pipeline automatically we were doing it with a random generator, which prevents the cache from working. This changes the behaviour to be...
## Description This PR adds a new mixin that contains the `_create_signature` method for the pipeline, and adds a new attribute named `exclude_from_signature` which consists of a `set` of strings...
## Description As part of issue https://github.com/argilla-io/distilabel/issues/832, we started adding default structured outputs for tasks that could benefit from it. `InstructionBacktranslation` is a good candidate. PR https://github.com/argilla-io/distilabel/pull/868 can be used...
## Description As part of issue https://github.com/argilla-io/distilabel/issues/832, we started adding default structured outputs for tasks that could benefit from it. `SelfInstruct` is a good candidate. PR https://github.com/argilla-io/distilabel/pull/868 can be used...