base_model: /root/cproject_updated/Qwen2.5-7B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
trust_remote_code: true

plugins:
  - axolotl.integrations.liger.LigerPlugin

liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: false

load_in_8bit: false
load_in_4bit: false
strict: false

auto_resume_from_checkpoints: true

datasets:
  - path: json
    data_files: /root/cproject_updated/judge_1k_axolotl.jsonl
    ds_type: json
    type: completion

shuffle_merged_datasets: true
dataset_prepared_path: /root/cproject_updated/prnew142
val_set_size: 0.05
output_dir: /root/cproject_updated/conv_200k_14b
sequence_len: 8192
sample_packing: true
eval_sample_packing: false

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_adamw_8bit
lr_scheduler: cosine

learning_rate: 1e-5
adam_beta1: 0.99
adam_beta2: 0.99
max_grad_norm: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: true

logging_steps: 1
flash_attention: true
warmup_steps: 10
eval_steps: 26
saves_per_epoch: 1

deepspeed: /sky_workdir/axolotl/deepspeed_configs/zero3_bf16.json

auto_resume_from_checkpoints: false
wandb_project: corruption_model_rm
wandb_entity:
wandb_watch:
wandb_name: rm-test-v1-7b-adammax2
wandb_log_model:

root/cproject_updated/conv_200k_14b

This model was trained from scratch on the json dataset. It achieves the following results on the evaluation set:

Loss: 0.5430

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: 1e-05
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 8
total_eval_batch_size: 8
optimizer: Use paged_adamw_8bit with betas=(0.99,0.99) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 10
num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.7452	1.0	26	0.7888
0.6347	2.0	52	0.6729
0.6479	3.0	78	0.5560
0.3729	4.0	104	0.5430

Framework versions

Transformers 4.47.1
Pytorch 2.4.1+cu121
Datasets 3.1.0
Tokenizers 0.21.0

Quest-AI/rm-proto-7b-s104-v1

root/cproject_updated/conv_200k_14b

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Quest-AI/
rm-proto-7b-s104-v1