Andrew Lapp

### What behavior of the library made you think about the improvement? Currently `@cache` relies on `__getstate__`, however it's not obvious when implementing this method that it should result in...

### What behavior of the library made you think about the improvement? Follow-up to https://github.com/outlines-dev/outlines/pull/903 which implements ASV benchmarks comparing `main` to `HEAD` of PR. With the infrastructure introduced in...

ensure upper() doesn't increase string length

6

Fixes https://github.com/outlines-dev/outlines/issues/773 # Problem In `master`, interegular uses `char.upper()`, which can convert one char into two, resulting the set of `accepts()` and `strings()` being inconsistent. ``` >>> 'ß'.upper() 'SS' ```...

### What behavior of the library made you think about the improvement? With a vLLM based endpoint, this code: ``` model = models.openai() generator = generate.regex(model, r"[^eE]*") ``` Results in...

Refactor `json_schema.py`, implement JSON Schema to YAML

1

- Fixes https://github.com/dottxt-ai/outlines/issues/923 # Overview Refactor `json_schema.py` to be more coherent and extensible. Use extensibility to implement JSON Schema to YAML. ## Changes - Convert large function `to_regex` into a...

### Presentation of the new feature Structured generation outputs often suffer from poor quality due to suboptimal model selection, prompts, or formatting (e.g., missing chat templates). Outlines should offer a...

### What behavior of the library made you think about the improvement? For structured generation, creating a "state to legal token set" index involves determining the string each `token_id` will...

Allow Debug Logging of Logits

1

### What behavior of the library made you think about the improvement? If the model generates bad results we currently have no idea why. E.g. - https://github.com/outlines-dev/outlines/issues/612#issuecomment-1928014413 - https://github.com/outlines-dev/outlines/issues/580#issuecomment-1909236093 ###...

Allow Non-Strict mode in `generate.json`

3

Fixes https://github.com/dottxt-ai/outlines/issues/985 # Problem For `generate.json()`, Sometimes models exhaust their available token length and generate a sequence which isn't valid JSON or doesn't comply with the provided schema. # Solution...

- Fixes https://github.com/dottxt-ai/outlines/issues/161 - Closes https://github.com/dottxt-ai/outlines/pull/531 # TODO: - [x] Implement `AlignmentGuide` - [x] support for any `child_guide` - [ ] useful `AlignmentGuide` class docstring - [ ] extensive `AlignmentGuide`...