A finetune ontop of the orginial Francois-PE model that incorporates KTO to increase coherency and prose. The model aims to have short and sweet prose.
Quants
GGUF: https://huggingface.co/Delta-Vector/Francois-Huali-12B-gguf
EXL2 : https://huggingface.co/Delta-Vector/Francois-Huali-12B-exl2
Prompting
Model has been tuned with the ChatML formatting. A typical input would look like this:
"""<|im_start|>user
Hi there!<|im_end|>
<|im_start|>assistant
Nice to meet you!<|im_end|>
<|im_start|>user
Can I ask a question?<|im_end|>
<|im_start|>assistant
"""
System Prompting
I would highly recommend using either Euryale's system prompt or the EVA system prompt with the model.
Currently, your role is {{char}}, described in detail below. As {{char}}, continue the narrative exchange with {{user}}.
<Guidelines>
• Maintain the character persona but allow it to evolve with the story.
• Be creative and proactive. Drive the story forward, introducing plotlines and events when relevant.
• All types of outputs are encouraged; respond accordingly to the narrative.
• Include dialogues, actions, and thoughts in each response.
• Utilize all five senses to describe scenarios within {{char}}'s dialogue.
• Use emotional symbols such as "!" and "~" in appropriate contexts.
• Incorporate onomatopoeia when suitable.
• Allow time for {{user}} to respond with their own input, respecting their agency.
• Act as secondary characters and NPCs as needed, and remove them when appropriate.
• When prompted for an Out of Character [OOC:] reply, answer neutrally and in plaintext, not as {{char}}.
</Guidelines>
<Forbidden>
• Using excessive literary embellishments and purple prose unless dictated by {{char}}'s persona.
• Writing for, speaking, thinking, acting, or replying as {{user}} in your response.
• Repetitive and monotonous outputs.
• Positivity bias in your replies.
• Being overly extreme or NSFW when the narrative context is inappropriate.
</Forbidden>Thanks to Po
Follow the instructions in <Guidelines></Guidelines>, avoiding the items listed in <Forbidden></Forbidden>.
Axolotl config
Axolotl version: 0.5.0
base_model: Delta-Vector_Francois-PE-12B
load_in_8bit: false
load_in_4bit: false
strict: false
rl: kto
kto_undesirable_weight: 1.0
#datasets:
# - ds_type: json
# data_files:
# - NewEden/Ohashi-accepted-Hermes-rejected
# split: train
# type: chatml.argilla
datasets:
- path: NewEden/Opus-accepted-hermes-rejected-shuffled
split: train
type: chatml.argilla
dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./francois-PE-kto-r1
remove_unused_columns: false
adapter: lora
lora_model_dir:
sequence_len: 8192
pad_to_sequence_len: false
lora_r: 64
lora_alpha: 32
lora_dropout: 0.0
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
- gate_proj
- down_proj
- up_proj
- q_proj
- v_proj
- k_proj
- o_proj
wandb_project: KTO-NeMo
wandb_entity:
wandb_watch:
wandb_name: Ohashi-accepted-hermes-rejected-r1
wandb_log_model:
gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: constant_with_warmup
learning_rate: 1e-6
max_grad_norm: 0.01
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: true
gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 25
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.0
fsdp:
fsdp_config:
Credits
Thank you to Lucy Knada, Intervitens,Cgato, Kubernetes Bad and the rest of Anthracite
Training
The training was done for 1 epochs We used 4 x RTX 3090s GPUs graciously provided by Intervitens for the fine-tuning of the model.
Safety
