50M-MSA-Egyptian-v1

50M-MSA-Egyptian-v1 — Bidirectional MSA ↔ Egyptian Arabic

A 51.8M-parameter small language model that translates both ways between Modern Standard Arabic (الفصحى) and Egyptian colloquial Arabic (المصرية العامية). A single set of weights serves both directions; a direction-specific Arabic 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:

Egyptian→MSA scores a bit higher, as expected — MSA is more standardized, so a single reference captures more of the valid output space. The saved weights are the best checkpoint by validation loss (eval_loss=1.249, epoch 2 of 3).

Decoding note: use plain greedy. A repetition penalty (1.2) was tested and lowered BLEU by ~12 points in both directions, because Arabic legitimately repeats short particles that the penalty suppresses.

Example translations

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

MSA → Egyptian

Egyptian → MSA

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

Usage

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

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

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

SYS_TO_EGY = "أنت مترجم محترف. ترجم النص من اللغة العربية الفصحى إلى اللهجة المصرية العامية."
SYS_TO_MSA = "أنت مترجم محترف. ترجم النص من اللهجة المصرية العامية إلى اللغة العربية الفصحى."

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("ليكمل الله لك الخير مقابل ضيافتك الرائعة.", SYS_TO_EGY))
# → ربنا يكرمك بالخير على ضيافتك الحلوة.
print(translate("ربنا يجازيك خير على كرمك وحسن ضيافتك.", SYS_TO_MSA))
# → ليجلب الله لك الخير على كرمك، وحسن ضيافتك.

Training

• Base model: oddadmix/50M-2048-Emhotob (Llama arch, ~51.8M params)
• Dataset: oddadmix/egyptian-msa-2.9-openai-bytedance-translations (132K rows, egyptian/msa columns)
• Method: HuggingFace Trainer, ChatML, prompt-masked cross-entropy. Each row is exploded into two training examples (one per direction, ~258K 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 and occasional drift on long inputs. Idioms and honorifics are mostly handled well.
• Gender is disambiguated only from context; ambiguous inputs may default one way.
• Trained on conversational Egyptian ↔ MSA; other dialects are out of scope.

License

Apache-2.0, inherited from the base model.