See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: NousResearch/Yarn-Llama-2-7b-64k
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 31cd3565080df0c7_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/31cd3565080df0c7_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: output
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device: cuda
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 4
flash_attention: false
fp16: null
fsdp: null
fsdp_config:
  max_steps: 50
  weight_decay: 0.01
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: false
hub_model_id: tarabukinivan/fcf22042-5cbf-4b8a-a7dc-0b10ed28a3d4
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_memory:
  0: 70GiB
max_steps: 50
micro_batch_size: 2
mlflow_experiment_name: /tmp/31cd3565080df0c7_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_torch
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 70
sequence_len: 2048
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: fcf22042-5cbf-4b8a-a7dc-0b10ed28a3d4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: fcf22042-5cbf-4b8a-a7dc-0b10ed28a3d4
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

fcf22042-5cbf-4b8a-a7dc-0b10ed28a3d4

This model is a fine-tuned version of NousResearch/Yarn-Llama-2-7b-64k on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1372

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 50

Training results

Training Loss	Epoch	Step	Validation Loss
9.0167	0.0047	1	2.5160
7.9813	0.0234	5	2.4597
7.3818	0.0468	10	1.8492
4.5428	0.0702	15	0.8182
1.8211	0.0936	20	0.2940
1.0316	0.1170	25	0.2084
0.6347	0.1404	30	0.1485
0.5765	0.1637	35	0.1538
0.5892	0.1871	40	0.1426
0.5079	0.2105	45	0.1370
0.4535	0.2339	50	0.1372

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1