base_model: mistralai/Mixtral-8x7B-v0.1
model_type: MixtralForCausalLM
tokenizer_type: LlamaTokenizer
is_mistral_derived_model: false

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: /workspace/dataset/Aesir
    type: completion
  - path: /workspace/dataset/LimaRP
    type: completion
  - path: /workspace/dataset/no-robots-alpaca-completion-train.json
    type: completion
  - path: /workspace/dataset/toxicsharegpt-NoWarning-completion.jsonl
    type: completion

dataset_prepared_path:
val_set_size: 0
output_dir: /workspace/Noromaid-v0.1-mixtral-8x7b
resume_from_checkpoint: 
hf_use_auth_token:

adapter: qlora
lora_model_dir:

sequence_len: 16384
sample_packing: true
pad_to_sequence_len: true

lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - w1
  - w2
  - w3
lora_target_linear:
lora_fan_in_fan_out:
lora_modules_to_save:
  - embed_tokens
  - lm_head

wandb_project: Noromaid-v0.1-mixtral-8x7b-v3
wandb_entity: 
wandb_watch:
wandb_run_id: 
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 2
num_epochs: 3
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 0.0002

model_config:
  output_router_logits: true

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

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

loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_table_max_new_tokens: 128
saves_per_epoch: 2
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens: