Improve model card and add metadata

Hi! I'm Niels, part of the community science team at Hugging Face.

I noticed that this model card could be improved to help users better understand and use your work. This PR:
- Adds the `pipeline_tag: text-generation` to the metadata to improve discoverability.
- Adds a link to the associated paper: [LineageFlow: Flow Matching for High-Fidelity Family-Aware Protein Sequence Generation](https://huggingface.co/papers/2605.22252).
- Links the official GitHub repository.
- Provides sample usage instructions for downloading the checkpoint and running batch generation, as found in your repository's documentation.

Feel free to merge this if it looks good to you!

README.md

@@ -1,6 +1,7 @@
 ---
-license: mit
 library_name: pytorch
+license: mit
+pipeline_tag: text-generation
 tags:
 - protein-sequence-generation
 - flow-matching
@@ -11,27 +12,44 @@ tags:
 
 # LineageFlow RP55 Checkpoint
 
-This repository hosts the released LineageFlow checkpoint for family-aware protein sequence generation.
+LineageFlow is a Dirichlet flow-matching model designed for high-fidelity, family-aware protein sequence generation. It initializes generation from lineage priors derived from ancestral sequence reconstruction (ASR), turning generation into structured mutation from an evolved scaffold.
+
+- **Paper:** [LineageFlow: Flow Matching for High-Fidelity Family-Aware Protein Sequence Generation](https://huggingface.co/papers/2605.22252)
+- **Code:** [GitHub Repository](https://github.com/Jinx-byebye/LineageFlow)
 
-## Files
+## Model Description
 
-- `lineageflow-rp55.ckpt`: LineageFlow denoiser checkpoint.
-- `SHA256SUMS`: checksum for verifying the checkpoint download.
+Current discrete generative models for proteins often start from uniform or masked-token noise, which can discard position-specific constraints induced by evolution. LineageFlow addresses this by using phylogeny-informed priors to maintain family validity and structural confidence while exploring within-family diversity. Across diverse protein families, LineageFlow achieves family validity close to natural sequences and improves predicted structural confidence over uniform or mask-initialized baselines.
 
 ## Usage
 
+### Download Checkpoint
+
+You can download the checkpoint using the Hugging Face CLI:
+
 ```bash
+pip install -U "huggingface_hub[cli]"
+
 hf download jinxbye/LineageFlow \
   lineageflow-rp55.ckpt \
   --local-dir checkpoints
 ```
 
-See the GitHub repository for inference and evaluation code:
+### Batch Generation
+
+To generate a batch of sequences using the official inference script, run:
 
-```text
-https://github.com/jinxbye/LineageFlow
+```bash
+python inference/batch_generate.py \
+  --config config/generation.json \
+  --ckpt checkpoints/lineageflow-rp55.ckpt \
+  --num-samples 512 \
+  --gpus all \
+  --out outputs/lineageflow_samples.fasta
 ```
 
+For more detailed instructions on installation and single-family generation, please refer to the [GitHub repository](https://github.com/Jinx-byebye/LineageFlow).
+
 ## Citation
 
 ```bibtex
@@ -41,4 +59,4 @@ https://github.com/jinxbye/LineageFlow
   booktitle = {International Conference on Machine Learning},
   year      = {2026}
 }
-```
+```