See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: NousResearch/Yarn-Llama-2-13b-128k
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 337c10342c0a1f98_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/337c10342c0a1f98_train_data.json
  type:
    field_input: News Description
    field_instruction: Headline
    field_output: sentiment
    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
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: false
hub_model_id: fedovtt/cd11203b-5f42-468f-90fb-5f97b232714e
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0001
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: 75GiB
max_steps: 15
micro_batch_size: 2
mlflow_experiment_name: /tmp/337c10342c0a1f98_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: 5
sequence_len: 1024
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: cd11203b-5f42-468f-90fb-5f97b232714e
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: cd11203b-5f42-468f-90fb-5f97b232714e
warmup_steps: 5
weight_decay: 0.1
xformers_attention: true

cd11203b-5f42-468f-90fb-5f97b232714e

This model is a fine-tuned version of NousResearch/Yarn-Llama-2-13b-128k on the None dataset. It achieves the following results on the evaluation set:

• Loss: 4.0140

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: 0.0001
• 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: 5
• training_steps: 15

Training results

Training Loss	Epoch	Step	Validation Loss
45.9604	0.0004	1	11.4065
46.8527	0.0007	2	11.4026
48.3234	0.0015	4	11.2533
42.5346	0.0022	6	10.2569
35.7466	0.0030	8	7.8034
23.9333	0.0037	10	5.6048
18.2533	0.0045	12	4.3734
16.3781	0.0052	14	4.0140

Framework versions

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