Text Classification
Transformers
Safetensors
Hindi
English
bert
hinglish
code-mixed
research
hate-speech-detection
text-embeddings-inference
Instructions to use OmTheLast/muril-hinglish-mixed-cm-thar with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use OmTheLast/muril-hinglish-mixed-cm-thar with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="OmTheLast/muril-hinglish-mixed-cm-thar")# Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("OmTheLast/muril-hinglish-mixed-cm-thar") model = AutoModelForSequenceClassification.from_pretrained("OmTheLast/muril-hinglish-mixed-cm-thar", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download evaluation_summary.json from OmTheLast/muril-hinglish-mixed-cm-thar: direct link, hf CLI and curl.
- Browser
- Download file 1.72 kB
-
https://huggingface.co/OmTheLast/muril-hinglish-mixed-cm-thar/resolve/main/evaluation_summary.json
- Command line
-
hf download hf://OmTheLast/muril-hinglish-mixed-cm-thar/evaluation_summary.json
-
curl -L -o evaluation_summary.json https://huggingface.co/OmTheLast/muril-hinglish-mixed-cm-thar/resolve/main/evaluation_summary.json
1.72 kB
| { | |
| "evaluation_role": "Evaluation split also used for best-epoch selection; not an untouched final test.", | |
| "mixed_external_evaluation": [ | |
| { | |
| "accuracy": "0.6098326359832636", | |
| "f1_macro": "0.3788174139051332", | |
| "f1_positive": "0.0", | |
| "fn": "373", | |
| "fp": "0", | |
| "model": "muril", | |
| "recall_positive": "0.0", | |
| "test_dataset": "kaggle_hinglish_hate", | |
| "test_rows": "956", | |
| "tn": "583", | |
| "tp": "0", | |
| "train_dataset": "mixed_cm_plus_thar" | |
| }, | |
| { | |
| "accuracy": "0.6457831325301204", | |
| "f1_macro": "0.3923865300146413", | |
| "f1_positive": "0.0", | |
| "fn": "147", | |
| "fp": "0", | |
| "model": "muril", | |
| "recall_positive": "0.0", | |
| "test_dataset": "cm_splits_codemixed", | |
| "test_rows": "415", | |
| "tn": "268", | |
| "tp": "0", | |
| "train_dataset": "mixed_cm_plus_thar" | |
| }, | |
| { | |
| "accuracy": "0.5277056277056277", | |
| "f1_macro": "0.345423632757155", | |
| "f1_positive": "0.0", | |
| "fn": "1091", | |
| "fp": "0", | |
| "model": "muril", | |
| "recall_positive": "0.0", | |
| "test_dataset": "thar_religion", | |
| "test_rows": "2310", | |
| "tn": "1219", | |
| "tp": "0", | |
| "train_dataset": "mixed_cm_plus_thar" | |
| } | |
| ], | |
| "per_seed": [ | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.5548564687377113, | |
| "eval_f1_hate": 0.0, | |
| "eval_f1_macro": 0.35685381891755186, | |
| "eval_loss": 0.6871203780174255, | |
| "eval_precision_hate": 0.0, | |
| "eval_precision_macro": 0.27742823436885566, | |
| "eval_recall_hate": 0.0, | |
| "eval_recall_macro": 0.5, | |
| "eval_runtime": 6.7265, | |
| "eval_samples_per_second": 378.059, | |
| "eval_steps_per_second": 47.276, | |
| "seed": 42 | |
| } | |
| ] | |
| } | |