An experimental release.
zerofata/GeneticLemonade-Unleashed SFT+DPO QLora finetune.
This is a creative model intended to excel at character driven RP / ERP. It has not been tested or trained on adventure stories or any large amounts of creative writing.
This model is designed to provide longer, narrative heavy responses where characters are portrayed accurately and proactively.
Play with these, they are not the 'best' settings just a stable baseline.
Llama-3-Instruct-Names but you will need to uncheck "System same as user".
The model first went through SFT with a small synthetic dataset of 2.9 million tokens, approximately 750 conversations. Primarily RP data with small amounts of random instruct / assistant data and creative writing.
The model then went through DPO training using approx 1100 chosen examples from the SFT dataset that were of exceptional quality or showed verifiable instruction following. Rejected samples were generated using another Llama 3.3 finetune that is known for poor instruction following.
Neither are optimized for cost / performance efficiency, YMMV.
# ====================
# MODEL CONFIGURATION
# ====================
base_model: zerofata/L3.3-GeneticLemonade-Unleashed-70B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
special_tokens:
pad_token: "<|finetune_right_pad_id|>"
chat_template: llama3
# ====================
# DATASET CONFIGURATION
# ====================
datasets:
- path: ./dataset.jsonl
type: chat_template
split: train
chat_template_strategy: tokenizer
field_messages: messages
message_property_mappings:
role: role
content: content
roles:
user: ["user"]
assistant: ["assistant"]
system: ["system"]
test_datasets:
- path: ./validate_dataset.jsonl
type: chat_template
split: train
chat_template_strategy: tokenizer
field_messages: messages
message_property_mappings:
role: role
content: content
roles:
user: ["user"]
assistant: ["assistant"]
system: ["system"]
dataset_prepared_path:
train_on_inputs: false # Only train on assistant responses
# ====================
# QLORA CONFIGURATION
# ====================
adapter: qlora
load_in_4bit: true
lora_r: 64
lora_alpha: 128
lora_dropout: 0.1
lora_target_linear: true
# lora_modules_to_save: # Uncomment only if you added NEW tokens
# ====================
# TRAINING PARAMETERS
# ====================
num_epochs: 2
micro_batch_size: 4
gradient_accumulation_steps: 2
learning_rate: 1.5e-5
optimizer: paged_adamw_8bit
lr_scheduler: rex
warmup_ratio: 0.05
weight_decay: 0.01
max_grad_norm: 1.0
# ====================
# SEQUENCE & PACKING
# ====================
sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
# ====================
# HARDWARE OPTIMIZATIONS
# ====================
bf16: auto
flash_attention: true
gradient_checkpointing: true
# ====================
# EVALUATION & CHECKPOINTING
# ====================
evaluation_strategy: steps
eval_steps: 5
save_strategy: steps
save_steps: 5
save_total_limit: 5 # Keep best + last few checkpoints
load_best_model_at_end: true
metric_for_best_model: eval_loss
greater_is_better: false
early_stopping_patience: 5
# ====================
# LOGGING & OUTPUT
# ====================
output_dir: ./output_model
logging_steps: 2
save_safetensors: true
# ====================
# WANDB TRACKING
# ====================
wandb_project: project_name
# wandb_entity: your_entity
# wandb_name: your_run_name
# ====================
# MODEL CONFIGURATION
# ====================
base_model: ApocalypseParty/unleashed-fulldata30
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
special_tokens: {}
chat_template: tokenizer_default
# ====================
# RL/DPO CONFIGURATION
# ====================
rl: dpo
rl_beta: 0.07
# ====================
# DATASET CONFIGURATION
# ====================
datasets:
- path: ./dpo_cleaned-v3_deduplicated.jsonl
type: chat_template.default
field_messages: conversation
field_chosen: chosen
field_rejected: rejected
message_property_mappings:
role: role
content: content
roles:
system: ["system"]
user: ["user"]
assistant: ["assistant"]
dataset_prepared_path:
train_on_inputs: false # Only train on assistant responses
# ====================
# QLORA CONFIGURATION
# ====================
adapter: qlora
load_in_4bit: true
lora_r: 32
lora_alpha: 64
lora_dropout: 0.05
lora_target_linear: true
# lora_modules_to_save: # Uncomment only if you added NEW tokens
# ====================
# TRAINING PARAMETERS
# ====================
num_epochs: 1
micro_batch_size: 4
gradient_accumulation_steps: 2
learning_rate: 2e-6
optimizer: adamw_8bit
lr_scheduler: cosine
warmup_steps: 5
weight_decay: 0.01
max_grad_norm: 1.0
# ====================
# SEQUENCE CONFIGURATION
# ====================
sequence_len: 4096
pad_to_sequence_len: true
# ====================
# HARDWARE OPTIMIZATIONS
# ====================
bf16: auto
tf32: false
flash_attention: true
gradient_checkpointing: offload
deepspeed: deepspeed_configs/zero1.json
# ====================
# CHECKPOINTING
# ====================
save_steps: 10
save_total_limit: 10
load_best_model_at_end: true
metric_for_best_model: eval_loss
greater_is_better: false
# ====================
# LOGGING & OUTPUT
# ====================
output_dir: ./dpo_model
logging_steps: 2
save_safetensors: true
# ====================
# WANDB TRACKING
# ====================
wandb_project: project_name
# wandb_entity: your_entity
# wandb_name: your_run_name
Base model
zerofata/L3.3-GeneticLemonade-Unleashed-70B