See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_resume_from_checkpoints: false
base_model: katuni4ka/tiny-random-olmo-hf
bf16: false
chat_template: llama3
dataset_prepared_path: null
dataset_processes: 6
datasets:
- data_files:
  - 221da31169c149b4_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/221da31169c149b4_train_data.json
  type:
    field_instruction: article
    field_output: highlights
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 1000
eval_table_size: null
flash_attention: true
fp16: true
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: error577/bc5fbfc9-f1e4-4387-aa26-0da42ee81a33
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: 64
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: null
micro_batch_size: 8
mlflow_experiment_name: /tmp/221da31169c149b4_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 1000
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: 2ccb6882-99f8-4efd-a1e6-16fa51e91956
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 2ccb6882-99f8-4efd-a1e6-16fa51e91956
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

bc5fbfc9-f1e4-4387-aa26-0da42ee81a33

This model is a fine-tuned version of katuni4ka/tiny-random-olmo-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 10.6147

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_BNB 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: 30
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
10.8267	0.0001	1	10.8318
10.6865	0.0516	1000	10.6904
10.6779	0.1032	2000	10.6735
10.6701	0.1548	3000	10.6622
10.6584	0.2064	4000	10.6574
10.6639	0.2580	5000	10.6540
10.6565	0.3096	6000	10.6512
10.6328	0.3612	7000	10.6494
10.6499	0.4128	8000	10.6472
10.6341	0.4643	9000	10.6411
10.6441	0.5159	10000	10.6368
10.6524	0.5675	11000	10.6335
10.6661	0.6191	12000	10.6324
10.634	0.6707	13000	10.6303
10.6418	0.7223	14000	10.6288
10.6284	0.7739	15000	10.6276
10.6361	0.8255	16000	10.6266
10.6471	0.8771	17000	10.6255
10.6367	0.9287	18000	10.6251
10.641	0.9803	19000	10.6236
10.6258	1.0319	20000	10.6232
10.6364	1.0835	21000	10.6237
10.642	1.1351	22000	10.6221
10.6405	1.1867	23000	10.6216
10.6022	1.2383	24000	10.6214
10.6063	1.2899	25000	10.6210
10.6538	1.3415	26000	10.6204
10.6155	1.3930	27000	10.6197
10.651	1.4446	28000	10.6196
10.649	1.4962	29000	10.6187
10.637	1.5478	30000	10.6186
10.6229	1.5994	31000	10.6185
10.6273	1.6510	32000	10.6180
10.6075	1.7026	33000	10.6178
10.62	1.7542	34000	10.6176
10.6506	1.8058	35000	10.6171
10.6336	1.8574	36000	10.6172
10.6363	1.9090	37000	10.6171
10.6311	1.9606	38000	10.6166
10.6164	2.0122	39000	10.6162
10.6309	2.0638	40000	10.6158
10.6464	2.1154	41000	10.6162
10.6402	2.1670	42000	10.6157
10.6167	2.2186	43000	10.6154
10.6317	2.2701	44000	10.6155
10.6009	2.3217	45000	10.6153
10.6353	2.3733	46000	10.6152
10.6209	2.4249	47000	10.6151
10.5978	2.4765	48000	10.6151
10.6234	2.5281	49000	10.6150
10.6156	2.5797	50000	10.6149
10.6199	2.6313	51000	10.6150
10.6364	2.6829	52000	10.6149
10.6253	2.7345	53000	10.6148
10.6177	2.7861	54000	10.6147
10.6132	2.8377	55000	10.6148
10.6297	2.8893	56000	10.6147
10.615	2.9409	57000	10.6147

Framework versions

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