Token Classification
Transformers
Safetensors
Ancient Greek (to 1453)
char_bert_joint
ancient-greek
classical-philology
character-level
part-of-speech-tagging
lemmatization
dependency-parsing
custom_code
Instructions to use Ericu950/CharDiff-grc-tagger-parser with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Ericu950/CharDiff-grc-tagger-parser with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("token-classification", model="Ericu950/CharDiff-grc-tagger-parser", trust_remote_code=True)# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Ericu950/CharDiff-grc-tagger-parser", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "cfg": { | |
| "kfold_dir": "/nobackup/proj/disk/dionysus/shared/GreekCharBERT/treebanks/oga_repo/kfold", | |
| "attn": "sdpa", | |
| "T": 2048, | |
| "W": 384, | |
| "micro_batch": 2, | |
| "eval_micro": 8, | |
| "lr_enc": 0.0001, | |
| "lr_head": 0.0008, | |
| "llrd": 1.0, | |
| "wd": 0.01, | |
| "clip": 1.0, | |
| "warmup_frac": 0.1, | |
| "epochs": 60, | |
| "patience": 8, | |
| "pool": "mean", | |
| "head_dropout": 0.33, | |
| "parse_dropout": 0.33, | |
| "w_xpos": 1.0, | |
| "w_script": 1.0, | |
| "w_upos": 0.2, | |
| "w_flat": 1.0, | |
| "use_cap": true, | |
| "scalar_mix": true, | |
| "d_arc": 500, | |
| "d_rel": 150, | |
| "w_parse": 1.0, | |
| "log_every": 10, | |
| "ckpt": "/nobackup/proj/disk/dionysus/shared/GreekCharBERT/runs/gcb_doc_clean/best.pt", | |
| "name": "joint_docclean_f3_s0", | |
| "seed": 0, | |
| "fold": 3, | |
| "out_dir": "/nobackup/proj/disk/dionysus/shared/GreekCharBERT/parser_data/runs/joint_docclean_f3_s0" | |
| }, | |
| "tcfg": { | |
| "pool": "mean", | |
| "head_dropout": 0.33, | |
| "w_xpos": 1.0, | |
| "w_flat": 1.0, | |
| "w_script": 1.0, | |
| "w_upos": 0.2, | |
| "use_cap": true, | |
| "scalar_mix": true | |
| }, | |
| "pcfg": { | |
| "d_arc": 500, | |
| "d_rel": 150, | |
| "dropout": 0.33, | |
| "n_labels": 24 | |
| }, | |
| "pretrain_cfg": { | |
| "name": "gcb_doc_clean", | |
| "seed": 0, | |
| "d_model": 1024, | |
| "depth": 32, | |
| "char_window": 256, | |
| "attn": "flex", | |
| "compile": true, | |
| "qk_norm": true, | |
| "seq_len": 8192, | |
| "micro_batch": 4, | |
| "grad_accum": 2, | |
| "lr": 0.0003, | |
| "wd": 0.1, | |
| "clip": 1.0, | |
| "warmup_frac": 0.04, | |
| "decay_frac": 0.4, | |
| "tier_weights": [ | |
| 1.0, | |
| 1.0, | |
| 0.3 | |
| ], | |
| "lam": 0.1, | |
| "total_steps": 120000, | |
| "hard_max_steps": 1000000, | |
| "num_workers": 16, | |
| "log_every": 20, | |
| "ckpt_every": 2000, | |
| "eval_every_ckpt": true, | |
| "eval_n": 256, | |
| "out_dir": "/nobackup/proj/disk/dionysus/shared/GreekCharBERT/runs/gcb_doc_clean", | |
| "anneal_phases": [ | |
| [ | |
| 0.5, | |
| [ | |
| 1.0, | |
| 1.0, | |
| 0.3 | |
| ] | |
| ], | |
| [ | |
| 0.9, | |
| [ | |
| 1.0, | |
| 1.0, | |
| 0.0 | |
| ] | |
| ], | |
| [ | |
| 1.0, | |
| [ | |
| 1.0, | |
| 0.0, | |
| 0.0 | |
| ] | |
| ] | |
| ] | |
| }, | |
| "epoch": 55, | |
| "dev": { | |
| "epoch": 55, | |
| "xpos_exact": 0.9387, | |
| "lemma_script": 0.9406, | |
| "upos": 0.9739, | |
| "uas": 0.8982, | |
| "las": 0.848, | |
| "n_words": 182570, | |
| "score": 0.9253 | |
| }, | |
| "_source_checkpoint": "/nobackup/proj/disk/dionysus/shared/GreekCharBERT/parser_data/runs/joint_docclean_f3_s0/best.pt" | |
| } |