NEPSEE is a Nepali sentiment analysis benchmark created by having two annotators independently label 5,614 tweets from Nepali public figures on a five‑point scale. Only 3,189 sentences with identical polarity are kept for a three‑class version, while 766 sentences with adjacent labels are used in a five‑point task. Baseline results show NepBERTa achieving a macro‑F1 of 0.877 on the three‑class benchmark, with commercial LLMs performing similarly; on the five‑point task, Claude Opus 5.5 outperforms encoders by 0.10–0.12 macro‑F1, and a tokenizer audit reveals that restoring Devanagari vowel signs can significantly alter performance.
arXiv:2608.22922v1 Announce Type: new Abstract: We present HelaBERT, a family of two BERT-based masked language models pre-trained from scratch on approximately 1 billion tokens of Sinhala text sourc...
The paper audits a multilingual affective generation benchmark that uses emoji summaries for Bangla, English, and Hindi sentences. It finds that the benchmark’s conclusions are largely artifacts of the measurement instrument, with no system significantly outperforming another when annotators are treated as random factors. The study shows that annotator identity and output length drive most variance, and proposes a new stable metric called emoji‑affect decodability.
arXiv:2609.36194v1 Announce Type: new Abstract: Extracted sentiment directions can vary across samples even when downstream sentiment classification remains accurate. To evaluate direction reproducib...
The paper shows that HuggingFace’s ByteLevel pre‑tokenizer, which treats a word as a sequence of Unicode letters, splits abugida scripts at every vowel sign, creating a training‑free lower bound on tokenizer fertility. Across 26 languages, all 17 abugidas exhibit increased token counts (up to 9×), while Latin, Cyrillic, Hangul, and Han remain unchanged. The authors demonstrate that correcting the character class reduces Nepali token counts, improves model performance, and that this issue is widespread in popular HuggingFace models.
NE‑BERT is a multilingual encoder trained on about 8.3 million sentences from nine Northeast Indian languages plus Hindi and English. Using weighted sampling and a custom SentencePiece tokenizer, it achieves significantly lower perplexity than IndicBERT‑V2, MuRIL, and mBERT, and improves tokenization fertility. The model also addresses vocabulary fragmentation in extremely low‑resource languages through aggressive upsampling, and its effectiveness is validated on part‑of‑speech tagging for three of the languages.
arXiv:2609.37882v1 Announce Type: cross Abstract: Every text classifier for an African language begins with a budgeting question: how many labelled examples are needed, and can labels from other Afri...
arXiv:2608.30092v1 Announce Type: cross Abstract: We present Arkios, a 1.04B-parameter dense transformer pretrained from scratch on 150B tokens of bilingual English-Nepali text, using a custom single...
arXiv:2608. 07629v1 Announce Type: cross Abstract: Multilingual neural machine translation models such as NLLB-200 cover 200 languages but leave thousands unsupported, including most Grassfields Bantu languages of Cameroon.
arXiv:2512. 20757v2 Announce Type: replace-cross Abstract: Tokenizers provide the fundamental basis through which text is represented and processed by language models (LMs).
The paper compares generative and encoder-based neural models for multilingual Named Entity Recognition (NER) across the eleven languages of the Naamapadam benchmark. Five classic model families, four decoder-only large language models fine‑tuned with LoRA and 4‑bit NF4 quantisation, and nine generative models in zero‑to‑5‑shot inference were evaluated under strict CoNLL span‑level metrics. Encoder-based models (mBERT and XLM‑R) achieved substantially higher F1 scores—up to 0.675 on Hindi—than any generative architecture, with gaps of 7.5–40 percentage points; the best few‑shot result reached only 28% of the encoder baseline. The study identifies three language clusters (encoder‑dominant, partial‑coverage, and failure‑zone) and offers deployment guidelines based on transfer learning and low‑resource NLP principles.
arXiv:2507.23248v2 Announce Type: replace-cross Abstract: Bengali is spoken by more than 230 million people, yet no standardized instrument evaluates large language models (LLMs) on Bengali across th...
arXiv:2607. 23319v1 Announce Type: cross Abstract: Standard subword tokenization algorithms such as Byte-Pair Encoding (BPE) and SentencePiece are trained predominantly on modern language corpora and produce inefficient segmentations when applied to classical Indian languages.