Efe2898/caba-v1

ÇABA v1

A 50M-parameter recurrent causal language model continued from the pinned Kumru-based base checkpoint. This is a base pretrained model, not an instruction-tuned chat model.

Load with Transformers

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "Efe2898/caba-v1"
tokenizer = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, trust_remote_code=True, dtype=torch.float32
).eval()

The training/2xt4 directory records the latest continuation run. Original base-model files and training code are retained in this repository.