This is the many-to-English WMT21 competition model — a 4.7B-parameter dense encoder-decoder transformer trained for high-quality translation from 7 source languages into English. It won the WMT21 shared task for multiple language pairs. Architecture: 24 encoder + 24 decoder layers (d=2048, 32 heads, FFN=16384, ReLU, pre-norm, sinusoidal positions). Distributed under MIT license.

Files

File	Size	Notes
`wmt21-dense-24-wide-x-en-f16.gguf`	8.8 GB	F16 weights (reference quality)
`wmt21-dense-24-wide-x-en-q8_0.gguf`	4.7 GB	Q8_0 quantized (identical quality to F16 on test set)
`wmt21-dense-24-wide-x-en-q4_k.gguf`	2.5 GB	Q4_K quantized (minor word choice differences)

Supported languages (7 → English)

Code	Language
de	German
cs	Czech
ru	Russian
ja	Japanese
zh	Chinese
is	Icelandic
ha	Hausa

Target is always English (en).

Quick start

bash

# 1. Build CrispASR
git clone https://github.com/CrispStrobe/CrispASR
cd CrispASR
cmake -B build -DCMAKE_BUILD_TYPE=Release -DBUILD_SHARED_LIBS=OFF
cmake --build build -j

# 2. Pull model
huggingface-cli download cstr/wmt21-dense-24-wide-x-en-GGUF wmt21-dense-24-wide-x-en-q8_0.gguf --local-dir .

# 3. Translate
./build/bin/crispasr --backend m2m100 -m wmt21-dense-24-wide-x-en-q8_0.gguf \
    --text "Die Maschine lernt schnell und verändert die Welt." \
    -sl de -tl en

# Japanese → English
./build/bin/crispasr --backend m2m100 -m wmt21-dense-24-wide-x-en-q8_0.gguf \
    --text "機械学習は世界を変えています。" \
    -sl ja -tl en

Quality verification

Input	Language	F16	Q8_0	Q4_K
Hallo Welt, wie geht es dir heute?	de	Hello world, how are you today?	Hello world, how are you today?	Hello world, how are you today?
Die Maschine lernt schnell und verändert die Welt.	de	The machine learns quickly and changes the world.	The machine learns quickly and changes the world.	The machine learns quickly and changes the world.
機械学習は世界を変えています。	ja	Machine learning is changing the world.	Machine learning is changing the world.	Machine learning is changing the world.
Машинное обучение меняет мир.	ru	Machine learning is changing the world.	Machine learning is changing the world.	Machine learning is changing the world.

Architecture

Text → SentencePiece BPE tokenizer (128K vocab, 8 lang codes)
     → Source lang token (__de__) + text tokens + </s>
     → 24-layer transformer encoder (d=2048, 32 heads, FFN=16384, ReLU, pre-norm)
     → Sinusoidal positional embeddings (pre-computed)
     → 24-layer transformer decoder (self-attn + cross-attn + FFN)
     → Shared embedding LM head (tied weights)
     → English forced as first decoder token
     → Greedy decode → translated English text

Conversion

bash

python models/convert-m2m100-to-gguf.py \
    --input facebook/wmt21-dense-24-wide-x-en \
    --output wmt21-dense-24-wide-x-en-f16.gguf

Related models

cstr/wmt21-dense-24-wide-en-x-GGUF — English-to-many (reverse direction)
cstr/m2m100-418m-GGUF — smaller 100-language any-to-any model
facebook/wmt21-dense-24-wide-x-en — original PyTorch model

Provenance and EU AI Act Art. 53 note

Upstream model: facebook/wmt21-dense-24-wide-x-en — published by facebook.
Upstream licence: mit. This repository redistributes under the same terms; it grants no rights the upstream licence does not.
What was done here: format conversion and/or quantisation only (GGUF/GGML). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository. No training-content summary was found on the upstream model card at the time of writing; that documentation gap is upstream's and is not filled here.
Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.