Finnish fine-tunes
Collection
All my Finnish fine-tuned models.
•
23 items
•
Updated
•
2
LoRA trained with text-generation-webui in 4-bit using LumiOpen/Viking-7B as the base model for 1 epoch. Dataset used with the LoRA is pinzhenchen/alpaca-cleaned-fi.
It uses Alpaca format but with a translated instruction at the start:
{
"instruction,output": "Alla on ohje, jossa kuvataan tehtävä. Kirjoita vastaus, joka täyttää pyynnön asianmukaisesti.\n\n### Instruction:\n%instruction%\n\n### Response:\n%output%",
"instruction,input,output": "Alla on ohje, jossa kuvataan tehtävä ja joka on yhdistetty kontekstia lisäävään syötteeseen. Kirjoita vastaus, joka täyttää pyynnön asianmukaisesti.\n\n### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
}
Using the following settings:
{
"lora_name": "Viking-Alpaca-Finnish-Test-V1",
"always_override": false,
"q_proj_en": true,
"v_proj_en": true,
"k_proj_en": false,
"o_proj_en": false,
"gate_proj_en": false,
"down_proj_en": false,
"up_proj_en": false,
"save_steps": 500,
"micro_batch_size": 4,
"batch_size": 128,
"epochs": 1,
"learning_rate": "3e-4",
"lr_scheduler_type": "linear",
"lora_rank": 128,
"lora_alpha": 256,
"lora_dropout": 0.05,
"cutoff_len": 512,
"dataset": "alpaca_data_cleaned.fi",
"eval_dataset": "None",
"format": "alpaca-format-finnish",
"eval_steps": 100,
"raw_text_file": "None",
"overlap_len": 128,
"newline_favor_len": 128,
"higher_rank_limit": false,
"warmup_steps": 100,
"optimizer": "adamw_torch",
"hard_cut_string": "\\n\\n\\n",
"train_only_after": "",
"stop_at_loss": 0,
"add_eos_token": false,
"min_chars": 0,
"report_to": "None"
}
Base model
LumiOpen/Viking-7B