See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: trl-internal-testing/tiny-random-LlamaForCausalLM
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 8f8f0a0e97a4a8bc_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/8f8f0a0e97a4a8bc_train_data.json
  type:
    field_instruction: query
    field_output: response
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 30
eval_max_new_tokens: 128
eval_steps: 50
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 16
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/e2484bfd-a6e9-445d-be18-5176c4e49dc5
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: true
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
micro_batch_size: 4
mlflow_experiment_name: /tmp/8f8f0a0e97a4a8bc_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
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: 100
sequence_len: 2048
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: b2bc75f3-3043-4268-ab22-eefb45e66799
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: b2bc75f3-3043-4268-ab22-eefb45e66799
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

e2484bfd-a6e9-445d-be18-5176c4e49dc5

This model is a fine-tuned version of trl-internal-testing/tiny-random-LlamaForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 10.2666

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• 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: 10
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
10.3714	0.0003	1	10.3745
10.354	0.0146	50	10.3522
10.3303	0.0291	100	10.3285
10.3117	0.0437	150	10.3099
10.3007	0.0582	200	10.2985
10.2963	0.0728	250	10.2930
10.2904	0.0873	300	10.2897
10.288	0.1019	350	10.2874
10.2913	0.1164	400	10.2857
10.2867	0.1310	450	10.2842
10.2891	0.1455	500	10.2829
10.2841	0.1601	550	10.2817
10.2841	0.1746	600	10.2806
10.2826	0.1892	650	10.2795
10.2841	0.2037	700	10.2786
10.2816	0.2183	750	10.2778
10.2786	0.2328	800	10.2771
10.2794	0.2474	850	10.2764
10.2792	0.2619	900	10.2758
10.281	0.2765	950	10.2752
10.2762	0.2910	1000	10.2746
10.2771	0.3056	1050	10.2741
10.2781	0.3201	1100	10.2736
10.2764	0.3347	1150	10.2731
10.2761	0.3492	1200	10.2728
10.2756	0.3638	1250	10.2725
10.2735	0.3783	1300	10.2720
10.2736	0.3929	1350	10.2717
10.2762	0.4074	1400	10.2714
10.2785	0.4220	1450	10.2711
10.2759	0.4365	1500	10.2708
10.273	0.4511	1550	10.2705
10.2726	0.4656	1600	10.2703
10.2713	0.4802	1650	10.2700
10.2747	0.4947	1700	10.2697
10.2723	0.5093	1750	10.2694
10.2737	0.5238	1800	10.2692
10.273	0.5384	1850	10.2690
10.2747	0.5529	1900	10.2688
10.2734	0.5675	1950	10.2686
10.2741	0.5821	2000	10.2683
10.2748	0.5966	2050	10.2681
10.2752	0.6112	2100	10.2680
10.2672	0.6257	2150	10.2679
10.2771	0.6403	2200	10.2678
10.2724	0.6548	2250	10.2676
10.269	0.6694	2300	10.2675
10.2687	0.6839	2350	10.2674
10.2721	0.6985	2400	10.2673
10.2716	0.7130	2450	10.2672
10.2709	0.7276	2500	10.2671
10.274	0.7421	2550	10.2670
10.2669	0.7567	2600	10.2670
10.2703	0.7712	2650	10.2669
10.2682	0.7858	2700	10.2669
10.2697	0.8003	2750	10.2668
10.277	0.8149	2800	10.2668
10.272	0.8294	2850	10.2667
10.2723	0.8440	2900	10.2667
10.2654	0.8585	2950	10.2667
10.2698	0.8731	3000	10.2667
10.2719	0.8876	3050	10.2667
10.2731	0.9022	3100	10.2667
10.2691	0.9167	3150	10.2667
10.2731	0.9313	3200	10.2667
10.2708	0.9458	3250	10.2666
10.2708	0.9604	3300	10.2666
10.2707	0.9749	3350	10.2666
10.2682	0.9895	3400	10.2666

Framework versions

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