Add model

README.md

@@ -1,3 +1,47 @@
 ---
+language:
+ - en
+datasets:
+ - simpeval
+tags:
+ - simplification
 license: apache-2.0
 ---
+
+This contains the trained checkpoint for LENS-SALSA, as introduced in [**Dancing Between Success and Failure: Edit-level Simplification Evaluation using SALSA**](https://arxiv.org/abs/2305.14458). For more information, please refer to the [**SALSA repository**](https://github.com/davidheineman/salsa).
+
+```bash
+pip install lens-metric
+```
+
+```python
+from lens import download_model
+from lens.lens_salsa import LENS_SALSA
+
+model_path = download_model("davidheineman/lens-salsa")
+lens_salsa = LENS_SALSA(model_path)
+
+score = lens_salsa.score(
+ complex = [
+ "They are culturally akin to the coastal peoples of Papua New Guinea."
+ ],
+ simple = [
+ "They are culturally similar to the people of Papua New Guinea."
+ ]
+)
+```
+
+## Intended uses
+
+Our model is intented to be used for **reference-free simplification evaluation**. Given a source text and its translation, outputs a single score between 0 and 1 where 1 represents a perfect simplification and 0 a random simplification. LENS-SALSA was trained on edit annotations of the SimpEval dataset, which covers manually-written, complex Wikipedia simplifications. We have not evaluated our model on non-English languages or non-Wikipedia domains.
+
+## Cite SALSA
+If you find our paper, code or data helpful, please consider citing [**our work**](https://arxiv.org/abs/2305.14458):
+```tex
+@article{heineman2023dancing,
+ title={Dancing {B}etween {S}uccess and {F}ailure: {E}dit-level {S}implification {E}valuation using {SALSA}},
+ author = "Heineman, David and Dou, Yao and Xu, Wei",
+ journal={arXiv preprint arXiv:2305.14458},
+ year={2023}
+}
+```

checkpoints/epoch=3-step=1460-val_kendall=0.409.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:971a9c705c90bb97fe85e73211aa8ca2beff7e7f438395d2ac86403a4960c0b3
+size 1419010479

hparams.yaml

@@ -0,0 +1,40 @@
+activations: Tanh
+batch_size: 4
+class_identifier: unified_metric
+continuous_word_labels: false
+dropout: 0.15
+encoder_learning_rate: 1.0e-05
+encoder_model: RoBERTa
+final_activation: null
+hidden_sizes:
+- 384
+initalize_pretrained_unified_weights: true
+input_segments:
+- edit_id_simplified
+- edit_id_original
+keep_embeddings_frozen: true
+layer: mix
+layer_norm: true
+layer_transformation: sparsemax
+layerwise_decay: 0.95
+learning_rate: 3.1e-05
+load_pretrained_weights: true
+loss: mse
+loss_lambda: 0.9
+nr_frozen_epochs: 0.3
+optimizer: AdamW
+pool: avg
+pretrained_model: roberta-large
+score_target: lens_score
+sent_layer: mix
+span_targets:
+- edit_id_simplified
+- edit_id_original
+span_tokens:
+- bad
+warmup_steps: 0
+word_layer: 24
+word_level_training: true
+word_weights:
+- 0.1
+- 0.9