diff --git a/README.md b/README.md index 768e260..86fbfd8 100644 --- a/README.md +++ b/README.md @@ -1,8 +1,6 @@ # LLaMPPL + HuggingFace -
[![docs](https://github.com/probcomp/hfppl/actions/workflows/docs.yml/badge.svg)](https://probcomp.github.io/hfppl) -
LLaMPPL is a research prototype for language model probabilistic programming: specifying language generation tasks by writing probabilistic programs that combine calls to LLMs, symbolic program logic, and probabilistic conditioning. To solve these tasks, LLaMPPL uses a specialized sequential Monte Carlo inference algorithm. This technique, SMC steering, is described in [our recent workshop abstract](https://arxiv.org/abs/2306.03081). @@ -96,4 +94,4 @@ Further documentation can be found at https://probcomp.github.io/hfppl. ## Installation To get started, clone this repository and run `pip install .` -The example `examples/hard_constraints.py` uses LLaMA2, which requires authorization from Meta. To run it, create a file `examples/hf_api_token.py`, with a constant `HF_AUTH_TOKEN` set to your HuggingFace API token. Alternatively, you can switch to a different (open) model by changing the first argument of `CachedCausalLM.from_pretrained` to another HuggingFace model identifier. \ No newline at end of file +The example `examples/hard_constraints.py` uses LLaMA2, which requires authorization from Meta. To run it, create a file `examples/hf_api_token.py`, with a constant `HF_AUTH_TOKEN` set to your HuggingFace API token. Alternatively, you can switch to a different (open) model by changing the first argument of `CachedCausalLM.from_pretrained` to another HuggingFace model identifier. diff --git a/mkdocs.yml b/mkdocs.yml index 7ee3d27..9aef663 100644 --- a/mkdocs.yml +++ b/mkdocs.yml @@ -2,6 +2,8 @@ site_name: LLaMPPL Docs theme: name: "material" + palette: + scheme: slate markdown_extensions: - pymdownx.highlight: @@ -33,4 +35,4 @@ nav: - batching.md - caching.md - immutability.md - - Code Reference: reference/ \ No newline at end of file + - Code Reference: reference/