See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: teknium/OpenHermes-2.5-Mistral-7B
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 83d749d1d83f68a2_train_data.json
  ds_type: json
  field: prompt
  path: /workspace/input_data/
  split: train
  type: completion
ddp_find_unused_parameters: false
debug: null
deepspeed: null
early_stopping_patience: null
ema_decay: 0.995
ema_update_after_step: 200
eps: 1.0e-06
eval_max_new_tokens: 256
eval_table_size: null
evals_per_epoch: 4
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 1
gradient_checkpointing: true
gradient_clipping: 0.5
gradient_normalization: true
greater_is_better: false
group_by_length: false
hub_model_id: CheapsetZero/5a9022d0-f07f-44eb-abb9-166cd1900db0
learning_rate: 0.00018
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_nan_inf_filter: true
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 11220
metric_for_best_model: eval_loss
micro_batch_size: 24
min_lr: 3.6e-05
mlflow_experiment_name: /tmp/83d749d1d83f68a2_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
reward_model_sampling_temperature: 0.7
s2_attention: null
sample_packing: false
save_total_limit: 3
saves_per_epoch: 4
sequence_len: 1024
skip_nan_gradients: true
special_tokens:
  pad_token: <|im_end|>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trl:
  adaptive_beta: true
  beta: 0.12
  entropy_coeff: 0.01
  gradient_normalization: true
  kl_monitoring: true
  max_completion_length: 1024
  num_generations: 12
  reward_funcs:
  - rewards_1cd8f023-f9d4-4d99-93fe-b7d21bed195d.reward_short_sentences
  - rewards_1cd8f023-f9d4-4d99-93fe-b7d21bed195d.reward_low_unique_words_percentage
  - rewards_1cd8f023-f9d4-4d99-93fe-b7d21bed195d.reward_low_syllables_per_word
  - rewards_1cd8f023-f9d4-4d99-93fe-b7d21bed195d.reward_specific_char_count_normalized
  - rewards_1cd8f023-f9d4-4d99-93fe-b7d21bed195d.reward_high_syllables_per_word
  reward_weights:
  - 4.530131987296112
  - 0.5133070523414518
  - 3.9325955598986004
  - 5.0
  - 4.871272511499004
  target_kl: 0.01
  use_vllm: false
trust_remote_code: true
use_ema: true
use_peft: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: offline
wandb_name: 1cd8f023-f9d4-4d99-93fe-b7d21bed195d
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 1cd8f023-f9d4-4d99-93fe-b7d21bed195d
warmup_steps: 642
weight_decay: 0.01
xformers_attention: null

5a9022d0-f07f-44eb-abb9-166cd1900db0

This model is a fine-tuned version of teknium/OpenHermes-2.5-Mistral-7B on the None dataset. It achieves the following results on the evaluation set:

• Loss: nan

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: 0.00018
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• 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: 642
• training_steps: 768

Training results

Training Loss	Epoch	Step	Validation Loss
0.0	0.0039	1	nan
0.0	0.25	64	nan
0.0	0.5	128	nan
0.0	0.75	192	nan
0.0	1.0	256	nan
0.0	1.25	320	nan
0.0	1.5	384	nan
0.0	1.75	448	nan
0.0	2.0	512	nan
0.0	2.25	576	nan
0.0	2.5	640	nan
0.0	2.75	704	nan
0.0	3.0	768	nan

Framework versions

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