base_model: mistralai/Mistral-7B-Instruct-v0.3
# optionally might have model_type or tokenizer_type
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer
# Automatically upload checkpoint and final model to HF
hub_model_id: AiAF/QLoRA-Finetune-LMTV-TM-9-2320-365-10

load_in_8bit: false
load_in_4bit: true
#datasets:
#  - path: ./
#    data_files:
#      - simplified_data_no_rag_with_special_tokens.jsonl
#      - plain_qa_list_with_special_tokens.jsonl
     # - AiAF/TM_multi_turn_convs_DATAGEN_OUTPUT
#    ds_type: json
#    type: chat_template
#    chat_template: tokenizer_default #chatml
#    field_messages: conversations
#    message_field_role: from
#    message_field_content: value
#    roles:
#      user:
#        - human
#      assistant:
#        - gpt
#      system:
#        - system

datasets:
  - path: AiAF/TM-Datasets_Special-Tokens
    ds_type: json
    type: chat_template
    chat_template: tokenizer_default
    field_messages: conversations
    message_field_role: from
    message_field_content: value
    roles:
      user:
        - human
      assistant:
        - gpt
      system:
        - system

dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./TM-9-2320-365-10_V17

save_total_limit: 100

adapter: qlora
lora_model_dir:
lora_modules_to_save:
  - embed_tokens
  - lm_head
sequence_len: 2048
sample_packing: true
pad_to_sequence_len: true
lora_r: 256
lora_alpha: 512
lora_dropout: 0.05

lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj

wandb_project: "LLM-Pretraining"
wandb_watch: "all"
wandb_name: "LMTV-TM-V17"
wandb_log_model: "false"
wandb_run_id: "LMTV-TM-V17"

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 5
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

bf16: auto
tf32: false

gradient_checkpointing: true
resume_from_checkpoint:
logging_steps: 1
flash_attention: true

loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3


warmup_steps: 10
evals_per_epoch: 50
saves_per_epoch: 1
weight_decay: 0.0
special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
  pad_token: "<pad>"
  unk_token: "<unk>"

QLoRA-Finetune-LMTV-TM-9-2320-365-10

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.3 on the AiAF/TM-Datasets_Special-Tokens 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: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 10
num_epochs: 5.0

Training results

Framework versions

PEFT 0.15.1
Transformers 4.51.1
Pytorch 2.6.0+cu124
Datasets 3.5.0
Tokenizers 0.21.1