A translator between Toki Pona and English, Russian and
Vietnamese. Small enough to run on a phone: it powers
ilo toki, which does all of its
translation on device.
This repository holds both the merged weights and GGUF builds.
What changed in 1.4
Two things, and the second one matters more than the first.
Training.group_by_length is off. The datasets in the mix have different
length profiles, so grouping similar lengths into a batch also grouped by source;
mixing them was meant to improve the model's fit across the whole mixture. Measured
on the project's own held-out split it is a small, consistent gain: BLEU is up on
five of six directions against 1.3.
The merge is fixed, and this is the larger change. Releases 1.1 through 1.3
were merged with the output head untied from the trained embeddings, on the
belief that the head had never seen them during training. That belief was wrong for
the mechanism this adapter uses. The adapter trains individual embedding rows
through PEFT's trainable_token_indices, and PEFT explicitly finds the weights
tied to the embedding matrix and puts a tied adapter on each — so the output head
read the trained rows throughout training, and untying handed the merged model a
head it had never used.
Merging correctly changes 119 of 153 probe answers and fixes most of what earlier
releases were known for:
1.3 as released
1.4
toki pona named as another language
2 of 18 probes
0
jan li lape ala
«Someone is sleeping»
«People are awake»
jan li lape ala → Russian
«Люди спят не спится»
«Люди не спят»
la where the relation is causal
«If my computer breaks down…»
«…because my phone is broken»
unseen proper names
copied in latin script
tokiponized, phonotactically legal
It also makes the files about 23% smaller: llama.cpp stores one embedding
matrix for a tied model and two for an untied one, so Q8_0 is 0.996 GiB where 1.3
was 1.29.
If you are fine-tuning this base yourself with trainable_token_indices, do not
untie the head at merge time. The check is one line: build the peft model and look
for TrainableTokensLayer — there should be two, on embed_tokens and on
lm_head.
Prompt format
The model keeps the prompt format of its base, and there is no chat template —
do not wrap the input in one.
Translate this from Toki Pona to English:
Toki Pona: jan li moku e kili
English:
The translation follows the final <target language>: line and ends at the model's
end-of-generation token. Either side can be the source:
Translate this from Russian to Toki Pona:
Russian: Я тебя люблю.
Toki Pona:
Language names are written out in full — Toki Pona, English, Russian,
Vietnamese. Getting the format wrong does not fail loudly: the model keeps
producing fluent text while silently ignoring the requested target language.
Which file to use
File
Size
Notes
ilo-toki-1.4-MiLMMT-46-1b-Q4_K_M.gguf
0.81 GB
Smallest.
ilo-toki-1.4-MiLMMT-46-1b-Q5_K_M.gguf
0.85 GB
ilo-toki-1.4-MiLMMT-46-1b-Q6_K.gguf
1.01 GB
ilo-toki-1.4-MiLMMT-46-1b-Q8_0.gguf
1.07 GB
What the app ships.
model.safetensors
2.00 GB
Merged weights, bf16, for transformers.
Running it
shell
1llama-completion -m ilo-toki-1.4-MiLMMT-46-1b-Q8_0.gguf --temp 0 --top-k 1 -no-cnv \2 -p "Translate this from Toki Pona to English:
3Toki Pona: jan li moku e kili
4English:"
Greedy decoding is what this model is meant to be run with. There is one right
answer per input, and sampling only ever walks away from it.
How it was built
A LoRA adapter (NetherQuartz/ilo-toki-1.4-MiLMMT-46-1b,
revision 48f57ca984) trained with TRL SFT — rank 64, targeting the attention and
MLP projections, plus 15 764 individual embedding rows through PEFT's
trainable_token_indices — merged into
MiLMMT-46-1B-v0.1 with
the output head left tied, and quantized with llama.cpp. The checkpoint is the one
at 17 500 steps, the minimum of the validation loss.
Alongside tok↔x pairs the mix includes x↔y pairs between the natural
languages, meant to keep their generation fluent without crowding out the pairs
where Toki Pona is one side.
Known limitations
Short inputs still acquire invented specifics.soweli lili li lape lon tomo
into Russian can produce a mouse that the sentence never mentioned.
Unmarked features get a default rather than a reading. Toki Pona marks
neither number nor tense; both readings are valid, but the model picks rather
than infers from context.
Transliteration of unseen names is legal but not conventional.Fukuoka
comes back as a well-formed toki pona word that is not the attested one. Names
the training data has seen are right.
