Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-2250/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,66 @@
+---
+license: mit
+base_model: FacebookAI/xlm-roberta-large
+tags:
+- token-classification
+- relation-extraction
+- romanian
+- span-detection
+- ner
+language:
+- en
+library_name: transformers
+pipeline_tag: token-classification
+---
+
+# Span detector (English) — entity detection for end-to-end RE
+
+[Dragoș Mitruț Vasile](https://scholar.google.com/citations?user=eD-SutAAAAAJ) · [Elena-Simona Apostol](https://scholar.google.com/citations?user=XUZcjpEAAAAJ) · [Stefan-Adrian Toma](https://scholar.google.com/citations?user=wsz8cUgAAAAJ) · [Adrian Paschke](https://scholar.google.com/citations?user=D_ZARycAAAAJ) · [Ciprian-Octavian Truică](https://scholar.google.com/citations?user=ZOKqr-QAAAAJ)
+
+[![Paper](https://img.shields.io/badge/Paper-InProgress-blue)](https://When-Paper-Appears-it-Will-Work.com)
+[![arXiv](https://img.shields.io/badge/arXiv-WIP-b31b1b)](https://arxiv.org/abs/WIP)
+[![Website](https://img.shields.io/badge/Project-Website-green)](https://github.com/DS4AI-UPB/crosslingual-romanian-re)
+[![GitHub](https://img.shields.io/badge/GitHub-Code-black?logo=github)](https://github.com/DS4AI-UPB/crosslingual-romanian-re)
+[![License](https://img.shields.io/badge/License-CC%20BY--NC--SA%204.0-lightgrey)](https://creativecommons.org/licenses/by-nc-sa/4.0/)
+
+A token-classification model (`FacebookAI/xlm-roberta-large`, 560M) that tags the two entity spans `e1` and `e2` in a plain English sentence, using a five-label BIO scheme (`O`, `B-E1`, `I-E1`, `B-E2`, `I-E2`). It is the first stage of the **end-to-end pipeline baseline**: its predicted spans are wrapped in entity markers and passed to the relation classifier ([`DS4AI-UPB/xlmr-large-ro-re`](https://huggingface.co/DS4AI-UPB/xlmr-large-ro-re)).
+
+The SemEval-2010 Task 8 entities are common nominals, not named entities, so an off-the-shelf NER model does not transfer — this detector is trained on the task's own spans instead.
+
+## Results (span F1-Score, held-out validation split)
+
+| Metric | Value |
+|--------|-------|
+| Span F1-Score | 0.847 |
+| Precision | 0.83 |
+| Recall | 0.87 |
+
+## Training
+
+- Base: `FacebookAI/xlm-roberta-large`
+- Token classification, 5 BIO labels
+- 5 epochs, batch size 16, LR 2e-5, 10% warmup, weight decay 0.01
+- Best checkpoint by span F1-Score on a 10% validation split
+- Single NVIDIA A100 40GB, about 6 min
+
+## Usage
+
+```python
+from transformers import AutoModelForTokenClassification, AutoTokenizer
+
+tok = AutoTokenizer.from_pretrained("DS4AI-UPB/span-detector-en")
+model = AutoModelForTokenClassification.from_pretrained("DS4AI-UPB/span-detector-en")
+```
+
+The full pipeline (span detection then relation classification) is in `pipeline_e2e.py` in the [code repository](https://github.com/DS4AI-UPB/crosslingual-romanian-re).
+
+## Citation
+
+```bibtex
+@misc{vasile2026crosslingual,
+  title  = {Cross-lingual Relation Extraction with Large Language Models: Zero-Shot, Few-Shot, and Fine-Tuned Evaluation on Romanian},
+  author = {Vasile, Drago\c{s}-Mitru\c{t} and Apostol, Elena-Simona and Toma, \c{S}tefan-Adrian and Paschke, Adrian and Truic\u{a}, Ciprian-Octavian},
+  year   = {2026},
+  note   = {Preprint}
+}
+```

checkpoint-2250/config.json

@@ -0,0 +1,44 @@
+{
+  "add_cross_attention": false,
+  "architectures": [
+    "XLMRobertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
+  "id2label": {
+    "0": "O",
+    "1": "B-E1",
+    "2": "I-E1",
+    "3": "B-E2",
+    "4": "I-E2"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "is_decoder": false,
+  "label2id": {
+    "B-E1": 1,
+    "B-E2": 3,
+    "I-E1": 2,
+    "I-E2": 4,
+    "O": 0
+  },
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "xlm-roberta",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "output_past": true,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "type_vocab_size": 1,
+  "use_cache": false,
+  "vocab_size": 250002
+}

checkpoint-2250/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0004cd5c50a5b66f343e8d791270946147f17e7755ffc7e30b10e7f448df84d1
+size 2235432300

checkpoint-2250/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b539e96e23bd363a417321dadf040b2cd0cda692576431d6009574193d1fe4d
+size 4471109808

checkpoint-2250/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:970e1a1ba5e54e4ede549c88456284062383775387a6ec93b80ef62261c37157
+size 14645

checkpoint-2250/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4f2f32432556a40634d33d5c4f71ac519c3cc64e786917b630d15fbc1dc093d
+size 1465

checkpoint-2250/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19ffc61f7cb57aa8c419fa9d2409c2af1abf4a92ba5a12c60a09cafa5a59c9b1
+size 17098184

checkpoint-2250/tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "cls_token": "<s>",
+  "eos_token": "</s>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "tokenizer_class": "XLMRobertaTokenizer",
+  "unk_token": "<unk>"
+}

checkpoint-2250/trainer_state.json

@@ -0,0 +1,404 @@
+{
+  "best_global_step": 2250,
+  "best_metric": 0.8691119691119692,
+  "best_model_checkpoint": "models/span-detector-en/checkpoint-2250",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 2250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 91.02240753173828,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 1.4920175170898438,
+      "step": 50
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 2.9558212757110596,
+      "learning_rate": 8.8e-06,
+      "loss": 0.517423095703125,
+      "step": 100
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 3.402139186859131,
+      "learning_rate": 1.3244444444444447e-05,
+      "loss": 0.342627067565918,
+      "step": 150
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 17.39611053466797,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 0.26056709289550783,
+      "step": 200
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 5.071574687957764,
+      "learning_rate": 1.9762962962962965e-05,
+      "loss": 0.22308229446411132,
+      "step": 250
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 4.316466331481934,
+      "learning_rate": 1.9269135802469135e-05,
+      "loss": 0.20261220932006835,
+      "step": 300
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 5.521117687225342,
+      "learning_rate": 1.877530864197531e-05,
+      "loss": 0.1886991310119629,
+      "step": 350
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 5.815828800201416,
+      "learning_rate": 1.8281481481481482e-05,
+      "loss": 0.16674188613891602,
+      "step": 400
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 4.5299882888793945,
+      "learning_rate": 1.7787654320987655e-05,
+      "loss": 0.16345125198364258,
+      "step": 450
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.13898934423923492,
+      "eval_runtime": 2.262,
+      "eval_samples_per_second": 353.674,
+      "eval_span_f1": 0.8043394033320419,
+      "eval_span_precision": 0.7772369898914264,
+      "eval_span_recall": 0.8334002408671216,
+      "eval_steps_per_second": 22.105,
+      "step": 450
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 6.108893394470215,
+      "learning_rate": 1.729382716049383e-05,
+      "loss": 0.13114701271057128,
+      "step": 500
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 4.100574970245361,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.11392547607421875,
+      "step": 550
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 3.039893388748169,
+      "learning_rate": 1.6306172839506175e-05,
+      "loss": 0.11778846740722657,
+      "step": 600
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 5.236042022705078,
+      "learning_rate": 1.5812345679012345e-05,
+      "loss": 0.13898660659790038,
+      "step": 650
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 28.931020736694336,
+      "learning_rate": 1.531851851851852e-05,
+      "loss": 0.12507149696350098,
+      "step": 700
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 3.843799352645874,
+      "learning_rate": 1.4824691358024692e-05,
+      "loss": 0.12188647270202636,
+      "step": 750
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 4.53892183303833,
+      "learning_rate": 1.4330864197530865e-05,
+      "loss": 0.12907770156860351,
+      "step": 800
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 3.8453285694122314,
+      "learning_rate": 1.3837037037037038e-05,
+      "loss": 0.10693228721618653,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 3.4621620178222656,
+      "learning_rate": 1.334320987654321e-05,
+      "loss": 0.12036109924316406,
+      "step": 900
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.12033544480800629,
+      "eval_runtime": 2.2776,
+      "eval_samples_per_second": 351.255,
+      "eval_span_f1": 0.8471042471042473,
+      "eval_span_precision": 0.8254326561324304,
+      "eval_span_recall": 0.8699444885011896,
+      "eval_steps_per_second": 21.953,
+      "step": 900
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 6.322509765625,
+      "learning_rate": 1.2849382716049383e-05,
+      "loss": 0.07529039859771729,
+      "step": 950
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 3.418290615081787,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 0.07295596599578857,
+      "step": 1000
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 4.168809413909912,
+      "learning_rate": 1.1861728395061728e-05,
+      "loss": 0.07213904857635497,
+      "step": 1050
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 7.8667988777160645,
+      "learning_rate": 1.1367901234567902e-05,
+      "loss": 0.07192722797393798,
+      "step": 1100
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 7.623446464538574,
+      "learning_rate": 1.0874074074074075e-05,
+      "loss": 0.06722517967224122,
+      "step": 1150
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 6.239801406860352,
+      "learning_rate": 1.0380246913580247e-05,
+      "loss": 0.07196783065795899,
+      "step": 1200
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 3.0218799114227295,
+      "learning_rate": 9.88641975308642e-06,
+      "loss": 0.06527298927307129,
+      "step": 1250
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 4.90587854385376,
+      "learning_rate": 9.392592592592593e-06,
+      "loss": 0.0671483850479126,
+      "step": 1300
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 6.28033447265625,
+      "learning_rate": 8.898765432098767e-06,
+      "loss": 0.08484127044677735,
+      "step": 1350
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.12750950455665588,
+      "eval_runtime": 2.2166,
+      "eval_samples_per_second": 360.909,
+      "eval_span_f1": 0.8465549348230912,
+      "eval_span_precision": 0.7978237978237979,
+      "eval_span_recall": 0.9016263387544625,
+      "eval_steps_per_second": 22.557,
+      "step": 1350
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 2.5074634552001953,
+      "learning_rate": 8.404938271604938e-06,
+      "loss": 0.04905274868011475,
+      "step": 1400
+    },
+    {
+      "epoch": 3.2222222222222223,
+      "grad_norm": 6.0911335945129395,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 0.042406258583068845,
+      "step": 1450
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 2.984830856323242,
+      "learning_rate": 7.417283950617284e-06,
+      "loss": 0.04139186382293701,
+      "step": 1500
+    },
+    {
+      "epoch": 3.4444444444444446,
+      "grad_norm": 4.371617794036865,
+      "learning_rate": 6.923456790123458e-06,
+      "loss": 0.04349579334259033,
+      "step": 1550
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 3.0147387981414795,
+      "learning_rate": 6.42962962962963e-06,
+      "loss": 0.04111259937286377,
+      "step": 1600
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 2.9704976081848145,
+      "learning_rate": 5.935802469135803e-06,
+      "loss": 0.04131685256958008,
+      "step": 1650
+    },
+    {
+      "epoch": 3.7777777777777777,
+      "grad_norm": 7.620437145233154,
+      "learning_rate": 5.441975308641976e-06,
+      "loss": 0.042846541404724124,
+      "step": 1700
+    },
+    {
+      "epoch": 3.888888888888889,
+      "grad_norm": 5.5054755210876465,
+      "learning_rate": 4.9481481481481485e-06,
+      "loss": 0.04484175682067871,
+      "step": 1750
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.9470939636230469,
+      "learning_rate": 4.454320987654322e-06,
+      "loss": 0.03779443264007568,
+      "step": 1800
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.13389411568641663,
+      "eval_runtime": 2.3798,
+      "eval_samples_per_second": 336.163,
+      "eval_span_f1": 0.8595516382448745,
+      "eval_span_precision": 0.8292051756007394,
+      "eval_span_recall": 0.8922036595067622,
+      "eval_steps_per_second": 21.01,
+      "step": 1800
+    },
+    {
+      "epoch": 4.111111111111111,
+      "grad_norm": 3.932535171508789,
+      "learning_rate": 3.960493827160494e-06,
+      "loss": 0.025300028324127196,
+      "step": 1850
+    },
+    {
+      "epoch": 4.222222222222222,
+      "grad_norm": 1.4799652099609375,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 0.027079813480377198,
+      "step": 1900
+    },
+    {
+      "epoch": 4.333333333333333,
+      "grad_norm": 0.48528411984443665,
+      "learning_rate": 2.9728395061728397e-06,
+      "loss": 0.02175631046295166,
+      "step": 1950
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 4.015340328216553,
+      "learning_rate": 2.4790123456790126e-06,
+      "loss": 0.03089595079421997,
+      "step": 2000
+    },
+    {
+      "epoch": 4.555555555555555,
+      "grad_norm": 2.1856842041015625,
+      "learning_rate": 1.985185185185185e-06,
+      "loss": 0.026017727851867675,
+      "step": 2050
+    },
+    {
+      "epoch": 4.666666666666667,
+      "grad_norm": 2.22531795501709,
+      "learning_rate": 1.491358024691358e-06,
+      "loss": 0.022645549774169924,
+      "step": 2100
+    },
+    {
+      "epoch": 4.777777777777778,
+      "grad_norm": 3.4707603454589844,
+      "learning_rate": 9.97530864197531e-07,
+      "loss": 0.024853086471557616,
+      "step": 2150
+    },
+    {
+      "epoch": 4.888888888888889,
+      "grad_norm": 0.565341591835022,
+      "learning_rate": 5.037037037037038e-07,
+      "loss": 0.028703060150146485,
+      "step": 2200
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 3.8629367351531982,
+      "learning_rate": 9.876543209876544e-09,
+      "loss": 0.02438181161880493,
+      "step": 2250
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.15517345070838928,
+      "eval_runtime": 2.2136,
+      "eval_samples_per_second": 361.396,
+      "eval_span_f1": 0.8691119691119692,
+      "eval_span_precision": 0.8475150602409639,
+      "eval_span_recall": 0.8918383518225039,
+      "eval_steps_per_second": 22.587,
+      "step": 2250
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 2250,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3244325766054048.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-2250/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d54d68660b8b4b11a97b12ca47c021a8884afc800f696f481af4c1face6e9812
+size 5201

config.json

@@ -0,0 +1,44 @@
+{
+  "add_cross_attention": false,
+  "architectures": [
+    "XLMRobertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "dtype": "float32",
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
+  "id2label": {
+    "0": "O",
+    "1": "B-E1",
+    "2": "I-E1",
+    "3": "B-E2",
+    "4": "I-E2"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "is_decoder": false,
+  "label2id": {
+    "B-E1": 1,
+    "B-E2": 3,
+    "I-E1": 2,
+    "I-E2": 4,
+    "O": 0
+  },
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "xlm-roberta",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "output_past": true,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "type_vocab_size": 1,
+  "use_cache": false,
+  "vocab_size": 250002
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0004cd5c50a5b66f343e8d791270946147f17e7755ffc7e30b10e7f448df84d1
+size 2235432300

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19ffc61f7cb57aa8c419fa9d2409c2af1abf4a92ba5a12c60a09cafa5a59c9b1
+size 17098184

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "cls_token": "<s>",
+  "eos_token": "</s>",
+  "is_local": false,
+  "mask_token": "<mask>",
+  "model_max_length": 512,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "tokenizer_class": "XLMRobertaTokenizer",
+  "unk_token": "<unk>"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d54d68660b8b4b11a97b12ca47c021a8884afc800f696f481af4c1face6e9812
+size 5201