test_patches

base_model: model/reddit_4_merged
model_type: MistralForCausalLM
# tokenizer_use_mistral_common: true
tokenizer_type: LlamaTokenizer
tokenizer_kwargs:
    pretrained_model_name_or_path: mistralai/Mistral-7B-v0.1
    trust_remote_code: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: dataset/boolq.parquet
    split: train
    type:
      system_prompt: ""
      field_instruction: instruction
      field_output: answer
      format: "{instruction}"
      no_input_format: "{instruction}"
dataloader_drop_last: true

output_dir: ./patches/boolq_1_4_testb_patience_5

sequence_len: 256
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false

adapter: lora
lora_model_dir:


lora_r: 8
lora_alpha: 16
lora_dropout: 0.1
lora_fan_in_fan_out:
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 4
num_epochs: 5
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 5e-6

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

loss_watchdog_threshold: 20.0
loss_watchdog_patience: 5

warmup_steps: 30
# evals_per_epoch: 4
eval_steps: 250
val_set_size: 0.1
early_stopping_patience: 5
load_best_model_at_end: True
# early_stopping_monitor: 'eval_loss'
eval_table_size:
eval_max_new_tokens: 256
# saves_per_epoch: 4
save_steps: 250
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_layer_norm: true
liger_fused_linear_cross_entropy: true