See axolotl config

axolotl version: 0.4.0

base_model: meta-llama/Meta-Llama-3-8B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: true
strict: false

hub_model_id: davanstrien/query-gen

datasets:
  - path: davanstrien/query-gen
    type: alpaca
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./lora-out

sequence_len: 1024
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: axolotl
wandb_entity:
wandb_watch:
wandb_name: query
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 10
num_epochs: 4
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
   pad_token: <|end_of_text|>

query-gen

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2679

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: 0.0002
• train_batch_size: 10
• eval_batch_size: 10
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 160
• total_eval_batch_size: 40
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
2.8337	0.0071	1	2.8390
1.414	0.2540	36	1.4018
1.3212	0.5079	72	1.3332
1.304	0.7619	108	1.3042
1.2874	1.0159	144	1.2900
1.229	1.2522	180	1.2835
1.2247	1.5062	216	1.2779
1.2362	1.7601	252	1.2708
1.2364	2.0141	288	1.2663
1.1734	2.2504	324	1.2691
1.1781	2.5044	360	1.2683
1.1995	2.7584	396	1.2658
1.1861	3.0123	432	1.2626
1.1332	3.2487	468	1.2680
1.1438	3.5026	504	1.2680
1.1553	3.7566	540	1.2679

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.1.2+cu118
• Datasets 2.19.1
• Tokenizers 0.19.1