See axolotl config

axolotl version: 0.4.1

base_model: meta-llama/Meta-Llama-3-70B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: afrias5/FinUpTagsNoTestNoExNew
    type: alpaca
    field: text

dataset_prepared_path: Allama3dataset
val_set_size: 0
output_dir: models/Allama370b
lora_model_dir: models/Allama370b/checkpoint-36
auto_resume_from_checkpoints: true
sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: False
adapter: lora
lora_r: 4
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_modules_to_save:
  - embed_tokens
  - lm_head

wandb_project: 'llama3run'
wandb_entity:
wandb_watch:
wandb_run_id:
wandb_name: 'A70b'                                       #change
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 8
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 0.0002

train_on_inputs: false
group_by_length: false
bf16: true
fp16:
tf32: false
hub_model_id: afrias5/Allama370b
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:
logging_steps: 1
warmup_steps: 10
# eval_steps: 300
saves_per_epoch: 1
save_total_limit: 12
debug:
deepspeed:
weight_decay: 0.0
fsdp:
deepspeed: deepspeed_configs/zero3_bf16.json
fsdp_config:
special_tokens:
   pad_token: <|end_of_text|>

Allama370b

This model is a fine-tuned version of meta-llama/Meta-Llama-3-70B on the None dataset.

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• total_eval_batch_size: 2
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 8

Training results

Framework versions

• PEFT 0.11.1
• Transformers 4.42.4
• Pytorch 2.2.2+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1