Translation
Transformers
ONNX
Safetensors
PyTorch
Khmer
blenderbot
text2text-generation
transliteration
romanization
romanized-khmer
khmer
character-level
int8
Eval Results (legacy)
Instructions to use samdei/samdei-type-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use samdei/samdei-type-base with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "translation" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # pip install "transformers<5.0.0" from transformers import pipeline pipe = pipeline("translation", model="samdei/samdei-type-base")# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForSeq2SeqLM tokenizer = AutoTokenizer.from_pretrained("samdei/samdei-type-base") model = AutoModelForSeq2SeqLM.from_pretrained("samdei/samdei-type-base", device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 1,159 Bytes
7a7ad8f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | Samdei Type Base
Copyright 2026 The Samdei Type Authors
Licensed under the Apache License, Version 2.0 (see LICENSE). This covers every file in this
repository: the model weights, the ONNX graphs, onnx/vocab.json, the channel model, rerank.json
and the language models.
Language models (lm/char5_nowiki.npz, lm/wordfull_nowiki.json.gz)
Character and word n-gram counts of Khmer text, built without Wikipedia, from:
- HPLT v2 (HPLT/HPLT2.0_cleaned, khm_Khmr): CC0 packaging; the Common Crawl Terms of Use
apply to the text.
- FineWeb-2 (HuggingFaceFW/fineweb-2, khm_Khmr): ODC-By 1.0; the Common Crawl Terms of Use
apply to the text.
- Asian Language Treebank (ALT), Khmer side: CC BY 4.0.
- Samdei Type chat sentences (samdei/samdei-type-resources): CC BY 4.0.
Training data
The model and the channel model were trained on samdei/samdei-type-pretraining-data and
samdei/samdei-type-finetuning-data (CC BY-SA 4.0); their cards list every source. Part of the
fine-tuning pairs were spelled with tables derived from the Chhunneng romanized Khmer
dataset, MIT License, Copyright (c) 2025 Chrea Chanchhunneng, and from Wikidata (CC0).
|