50M-English-MSA-v1

50M-English-MSA-v1 — Bidirectional English ↔ MSA (Modern Standard Arabic)

A 51.8M-parameter small language model that translates both ways between English and Modern Standard Arabic (الفصحى). A single set of weights serves both directions; a direction-specific system prompt selects which way to translate.

Finetuned from oddadmix/50M-2048-Emhotob, a tiny Arabic base model trained from scratch.

Evaluation

Evaluated on a deterministic held-out set of 3,000 pairs (seed=42), decoded greedily (do_sample=False, no repetition penalty), scored with sacreBLEU:

Scores are much higher than the dialectal (Egyptian) siblings of this model family: MSA is highly standardized, so a single reference captures most of the valid output space in both directions. The saved weights are the best checkpoint by validation loss (eval_loss=0.4945, epoch 2 of 3).

Decoding note: use plain greedy. A repetition penalty (1.2) was tested across these 50M translation models and lowered BLEU by 5–12 points, because Arabic legitimately repeats short particles that the penalty suppresses.

Example translations

Real greedy-decoded outputs from the held-out set:

English → MSA

MSA → English

A larger set of 20 examples per direction (with references) is in eval_bidirectional.json.

Usage

ChatML format. Pick the system prompt for the direction you want:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "oddadmix/50M-English-MSA-v1"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16).to("cuda").eval()

SYS_TO_MSA = "أنت مترجم محترف. ترجم النص الإنجليزي إلى اللغة العربية الفصحى."
SYS_TO_EN  = "You are a professional translator. Translate the Modern Standard Arabic text into English."

def translate(text: str, system: str) -> str:
    prompt = (
        f"<|im_start|>system\n{system}<|im_end|>\n"
        f"<|im_start|>user\n{text.strip()}<|im_end|>\n"
        f"<|im_start|>assistant\n"
    )
    ids = tok(prompt, return_tensors="pt", add_special_tokens=False).to(model.device)
    if tok.bos_token_id is not None:  # training prepends BOS
        bos = torch.tensor([[tok.bos_token_id]], device=model.device)
        ids["input_ids"] = torch.cat([bos, ids["input_ids"]], dim=1)
        ids["attention_mask"] = torch.cat([torch.ones_like(bos), ids["attention_mask"]], dim=1)
    out = model.generate(**ids, max_new_tokens=256, do_sample=False,
                         eos_token_id=tok.eos_token_id, pad_token_id=tok.pad_token_id)
    return tok.decode(out[0, ids["input_ids"].size(1):], skip_special_tokens=True).strip()

print(translate("Thank you very much, you are so kind.", SYS_TO_MSA))
# → شكرًا جزيلًا لك، أنت لطيف جدًا.
print(translate("أنا لست خائفًا من أي شيء حقًا.", SYS_TO_EN))
# → I'm not really afraid of anything.

Training

• Base model: oddadmix/50M-2048-Emhotob (Llama arch, ~51.8M params)
• Dataset: oddadmix/egyptian-msa-2.9-openai-bytedance-translations (132K rows; this model uses the english and msa columns)
• Method: HuggingFace Trainer, ChatML, prompt-masked cross-entropy (loss only on the assistant turn). Each row is exploded into two training examples (one per direction, ~254K total). Two ChatML special tokens (<|im_start|>, <|im_end|>) were added and embeddings resized.
• Hyperparameters: 3 epochs · effective batch 64 · LR 3e-4 (cosine, 5% warmup) · bf16 · max length 1024 · load_best_model_at_end on eval_loss.
• Split: 129,009 train / 3,000 deterministic held-out (seed=42), scored both directions.

Limitations

• A 50M model: expect errors on rare / technical vocabulary, proper nouns, and occasional drift or truncation on very long inputs. Everyday and conversational text is handled well.
• Gender is disambiguated only from context; ambiguous inputs may default one way.
• Trained on the MSA side of an Egyptian-sourced parallel corpus; highly formal, legal, or classical registers are out of scope. For dialectal Egyptian, see the sibling models oddadmix/50M-Egyptian-English-v1 and oddadmix/50M-MSA-Egyptian-v1.

License

Apache-2.0, inherited from the base model.