base_model: winglian/phi-1_5-2x-fib
base_model_config: winglian/phi-1_5-2x-fib
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
is_llama_derived_model: false
trust_remote_code: true
                                                                       
load_in_8bit: false
load_in_4bit: false                                  
strict: false       
                                                                       
datasets:
  - path: nampdn-ai/tiny-lessons
    type: completion
    field: textbook         
  - path: nampdn-ai/tiny-textbooks
    type: completion        
    field: textbook         

dataset_prepared_path: last_run_prepared
val_set_size: 0.01
output_dir: ./phi-2x-out
hub_model_id: winglian/phi-2x-pt

sequence_len: 2048
sample_packing: false
pad_to_sequence_len: 
resize_token_embeddings_to_32x: true

adapter: 
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: phi-2x-pt-tiny
wandb_entity:
wandb_watch:
wandb_run_id:
wandb_log_model:

gradient_accumulation_steps: 2
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_torch
adam_beta2: 0.95
adam_epsilon: 0.00001
max_grad_norm: 1.0
lr_scheduler: cosine
learning_rate: 0.00003

train_on_inputs: false
group_by_length: false
bf16: true
fp16: false
tf32: true

gradient_checkpointing: 
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: 

warmup_steps: 1000
eval_steps: 0.05
save_steps: 0.10
do_bench_eval: true
bench_dataset: "pharaouk/dharma-1/dharma_1_full.json"
save_total_limit: 20
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
  bos_token: "<|endoftext|>"
  eos_token: "<|endoftext|>"
  unk_token: "<|endoftext|>"
  pad_token: "<|endoftext|>"