base_model: mistralai/Mistral-Large-Instruct-2407
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: Doctor-Shotgun/C2-Stheno
    type: sharegpt
    conversation: mistral
  - path: anthracite-org/kalo-opus-instruct-22k-no-refusal
    type: sharegpt
    conversation: mistral
  - path: anthracite-org/nopm_claude_writing_fixed
    type: sharegpt
    conversation: mistral
chat_template: inst

dataset_prepared_path: /workspace/data/datasets/tokenized
val_set_size: 0.00
output_dir: /workspace/data/fft-out

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:

wandb_project: 123b-magnum-fft
wandb_entity:
wandb_watch:
wandb_name: v1-attempt-1
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.000002

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

gradient_checkpointing: unsloth
#gradient_checkpointing_kwargs:
  #use_reentrant: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:

warmup_steps: 20
evals_per_epoch:
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 2
debug:
deepspeed: deepspeed_configs/zero3_bf16.json
weight_decay: 0.01
fsdp:
fsdp_config:
special_tokens: