You can deploy this model and start earning money today!

BangorAI/
ALMA-Cymraeg-13B-0.1

Name: ALMA-Cymraeg-13B-0.1
Availability: InStock

text-generation-inference

endpoints_compatible

Views

No views yet

ALMA-Cymraeg-13B

Fersiwn Gymraeg o fodel cyfieithu ALMA a ddisgrifir yn https://arxiv.org/abs/2309.11674.
This is a Welsh version of the ALMA LLM-based translation model.

Mae'r model LLM yn seiliedig ar Lama-2-13B, gyda hyfforddiant parhaus ar ddata Gymreig OSCAR-2301 am 3 Epoch ac yna hyfforddiant cywrain pellach ar ddata Cofnod y Cynulliad a ddarparir gan TechIaith.

Mae fersiwn cyflymach sydd wedi ei gywasgu i 4.0bpw er mwyn llwytho mewn cof GPU o 10GB ar gael yma.

Fformat Sgwrs

Mae'r hyfforddiant cywrain wedi defnyddio'r fformat canlynol ar gyfer trosi o'r Saesneg i'r Gymraeg (a'r naill ffordd i'r llall).

Cyfieithwch y testun Saesneg canlynol i'r Gymraeg.
### Saesneg:
{prompt}

### Cymraeg:

Esiampl

python

from transformers import AutoModelForCausalLM, AutoTokenizer

device = "cuda"

model = AutoModelForCausalLM.from_pretrained("BangorAI/ALMA-Cymraeg-13B-0.1", torch_dtype=torch.float16, load_in_8bit=True)
tokenizer = AutoTokenizer.from_pretrained("BangorAI/ALMA-Cymraeg-13B-0.1")

prompt = """Cyfieithwch y testun Saesneg canlynol i'r Gymraeg.
### Saesneg:
For the first time, GPs no longer have to physically print, sign and hand a green paper prescription form to the patient or wait for it to be taken to the pharmacy. Instead, the prescription is sent electronically from the surgery via the IT system to the patient’s chosen pharmacy - even without the patient needing to visit the surgery to pick up a repeat prescription form.

### Cymraeg:
"""

model_inputs = tokenizer([prompt], return_tensors="pt").to(device)

generated_ids = model.generate(**model_inputs,
                               eos_token_id=tokenizer.eos_token_id,
                               top_k=90,
                               top_p=1.0,
                               temperature=0.3,
                               repetition_penalty=1.2,
                               max_new_tokens=500,
                               do_sample=True)
print(tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0])

Hawlfraint

Mae'r model yn seiliedig ar Llama2 ac felly dan drwydded gan Meta.
Mae'r data Cofnod y Cynulliad dan drywdded Llywodraeth Agored.