See axolotl config

axolotl version: 0.4.1

base_model: google/gemma-2-2b
model_type: Gemma2ForCausalLM
tokenizer_type: AutoTokenizer
chat_template: gemma

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: flydust/Magpie-100k-Gemma2-9B
    type: sharegpt
    chat_template: gemma
dataset_prepared_path: last_run_prepared
val_set_size: 0.001
output_dir: axolotl_out/gemma-2-2b-magpie-gemma2-9b

sequence_len: 4096
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

wandb_project: SynDa
wandb_entity:
wandb_watch:
wandb_name: gemma-2-2b-magpie-gemma2-9b
wandb_log_model:
hub_model_id: flydust/gemma-2-2b-magpie-gemma2-9b

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: true
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
# Disable flash attention
# flash_attention: false
# sdp_attention: falses
eager_attention: true

warmup_ratio: 0.1
evals_per_epoch: 5
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

gemma-2-2b-magpie-gemma2-9b

This model is a fine-tuned version of google/gemma-2-2b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6998

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 79
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
4.7852	0.0023	1	1.1984
0.8091	0.2011	86	0.8370
0.7305	0.4022	172	0.7686
0.6761	0.6033	258	0.7394
0.6618	0.8044	344	0.7141
0.6197	1.0056	430	0.6983
0.5014	1.1932	516	0.7058
0.4924	1.3943	602	0.7018
0.4887	1.5954	688	0.6997
0.4696	1.7966	774	0.6998

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1