See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: unsloth/tinyllama
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 74fd83b58bc4ad47_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/74fd83b58bc4ad47_train_data.json
  type:
    field_input: conversation
    field_instruction: note
    field_output: summary
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 0.001
eval_max_new_tokens: 128
eval_steps: 40
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/c45ed0e9-140f-4c89-9882-409edc73abaa
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 100
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
micro_batch_size: 32
mlflow_experiment_name: /tmp/74fd83b58bc4ad47_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
s2_attention: null
sample_packing: false
save_steps: 40
saves_per_epoch: 0
sequence_len: 512
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: 21068da4-737c-49df-9240-0bd8ff25df8b
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 21068da4-737c-49df-9240-0bd8ff25df8b
warmup_ratio: 0.05
weight_decay: 0.0
xformers_attention: null

c45ed0e9-140f-4c89-9882-409edc73abaa

This model is a fine-tuned version of unsloth/tinyllama on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1196

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: 0.0003
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_bnb_8bit 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: 1105
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0011	1	1.2239
No log	0.0452	40	1.1594
No log	0.0904	80	0.6742
0.9748	0.1356	120	0.2826
0.9748	0.1808	160	0.2384
0.2539	0.2260	200	0.2086
0.2539	0.2712	240	0.1876
0.2539	0.3164	280	0.1782
0.1888	0.3616	320	0.1701
0.1888	0.4068	360	0.1643
0.1653	0.4520	400	0.1599
0.1653	0.4972	440	0.1560
0.1653	0.5424	480	0.1529
0.1543	0.5876	520	0.1507
0.1543	0.6328	560	0.1481
0.1496	0.6780	600	0.1469
0.1496	0.7232	640	0.1446
0.1496	0.7684	680	0.1431
0.145	0.8136	720	0.1413
0.145	0.8588	760	0.1410
0.1417	0.9040	800	0.1388
0.1417	0.9492	840	0.1377
0.1417	0.9944	880	0.1363
0.1366	1.0395	920	0.1358
0.1366	1.0847	960	0.1354
0.1314	1.1299	1000	0.1333
0.1314	1.1751	1040	0.1335
0.1314	1.2203	1080	0.1319
0.1305	1.2655	1120	0.1324
0.1305	1.3107	1160	0.1303
0.1264	1.3559	1200	0.1289
0.1264	1.4011	1240	0.1278
0.1264	1.4463	1280	0.1272
0.1244	1.4915	1320	0.1267
0.1244	1.5367	1360	0.1249
0.1224	1.5819	1400	0.1244
0.1224	1.6271	1440	0.1261
0.1224	1.6723	1480	0.1239
0.1239	1.7175	1520	0.1227
0.1239	1.7627	1560	0.1229
0.121	1.8079	1600	0.1221
0.121	1.8531	1640	0.1208
0.121	1.8983	1680	0.1205
0.1186	1.9435	1720	0.1194
0.1186	1.9887	1760	0.1189
0.1117	2.0339	1800	0.1192
0.1117	2.0791	1840	0.1192
0.1117	2.1243	1880	0.1196

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1