trollek/danube2-1.8b-SlimOrcaDedup

h2o-danube2 with ChatML template

This model was first fine-tuned with BAdam on cgato/SlimOrcaDedupCleaned using LLama-Factory.

Template

<|im_start|>system
{{system}}<|im_end|>
<|im_start|>user
{{instruction}}<|im_end|>
<|im_start|>assistant
{{response}}<|im_end|>

BAdam config

### model
model_name_or_path: danube2-base-chatml

### method
stage: sft
do_train: true
finetuning_type: full
use_badam: true
badam_switch_mode: ascending
badam_switch_interval: 50
badam_verbose: 1
badam_start_block: 13
seed: 314

### dataset
dataset: slimorca_dedup_cleaned
template: hermes_chatml
cutoff_len: 8192
overwrite_cache: false
preprocessing_num_workers: 12

### output
output_dir: slim-chatml-badam
logging_steps: 5
save_steps: 1
save_strategy: epoch
plot_loss: true
overwrite_output_dir: false

### train
per_device_train_batch_size: 2
gradient_accumulation_steps: 4
learning_rate: 0.000005
num_train_epochs: 1
lr_scheduler_type: cosine
warmup_ratio: 0.01
bf16: true
flash_attn: fa2

### eval
val_size: 0.01
per_device_eval_batch_size: 1
eval_strategy: steps
eval_steps: 2000

BAdam training results

Training Loss	Epoch	Step	Validation Loss
0.8535	0.0889	2000	0.8340
0.8735	0.1778	4000	0.8128
0.8054	0.2668	6000	0.8008
0.7907	0.3557	8000	0.8002
0.8749	0.4446	10000	0.7972
0.7463	0.5335	12000	0.7899
0.7762	0.6225	14000	0.7870
0.8231	0.7114	16000	0.7854
0.8686	0.8003	18000	0.7801
0.9159	0.8892	20000	0.7877
0.8281	0.9782	22000	0.7786