Delta-Vector/Baldur-8B

An finetune of the L3.1 instruct distill done by Arcee, The intent of this model is to have differing prose then my other releases, in my testing it has achieved this and avoiding using common -isms frequently and has a differing flavor then my other models.

Quants

GGUF: https://huggingface.co./Delta-Vector/Baldur-8B-GGUF

EXL2: https://huggingface.co./Delta-Vector/Baldur-8B-EXL2

Prompting

Model has been Instruct tuned with the Llama-Instruct formatting. A typical input would look like this:

"""<|begin_of_text|><|start_header_id|>system<|end_header_id|>
You are an AI built to rid the world of bonds and journeys!<|eot_id|><|start_header_id|>user<|end_header_id|>
Bro i just wanna know what is 2+2?<|eot_id|><|start_header_id|>assistant<|end_header_id|>
"""

System Prompting

I would highly recommend using Sao10k's Euryale System prompt, But the "Roleplay Simple" system prompt provided within SillyTavern will work aswell.

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>

Follow the instructions in <Guidelines></Guidelines>, avoiding the items listed in <Forbidden></Forbidden>.

Axolotl config

See axolotl config

Axolotl version: 0.4.1

base_model: arcee-ai/Llama-3.1-SuperNova-Lite
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

#trust_remote_code: true

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: Gryphe/Sonnet3.5-SlimOrcaDedupCleaned
    type: chat_template
  - path: Nitral-AI/Cybersecurity-ShareGPT
    type: chat_template
  - path: Nitral-AI/Medical_Instruct-ShareGPT
    type: chat_template
  - path: Nitral-AI/Olympiad_Math-ShareGPT
    type: chat_template
  - path: anthracite-org/kalo_opus_misc_240827
    type: chat_template
  - path: NewEden/Claude-Instruct-5k
    type: chat_template
  - path: lodrick-the-lafted/kalo-opus-instruct-3k-filtered
    type: chat_template
  - path: anthracite-org/kalo-opus-instruct-22k-no-refusal
    type: chat_template
  - path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
    type: chat_template
  - path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
    type: chat_template
  - path: anthracite-org/kalo_misc_part2
    type: chat_template
  - path: Nitral-AI/Creative_Writing-ShareGPT
    type: chat_template
  - path: NewEden/Gryphe-Sonnet3.5-Charcard-Roleplay-unfiltered
    type: chat_template

chat_template: llama3
shuffle_merged_datasets: true
default_system_message: "You are an assistant that responds to the user."
dataset_prepared_path: prepared_dataset_memorycore
val_set_size: 0.0
output_dir: ./henbane-8b-r3

sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len:

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

wandb_project: henbane-8b-r3
wandb_entity:
wandb_watch:
wandb_name: henbane-8b-r3
wandb_log_model:

gradient_accumulation_steps: 32
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
#learning_rate: 3e-5
learning_rate: 1e-5

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

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 5
evals_per_epoch:
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 2
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero2.json
weight_decay: 0.05
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|finetune_right_pad_id|>
  eos_token: <|eot_id|>

Credits

Thank you to Lucy Knada, Kalomaze, Kubernetes Bad and the rest of Anthracite (But not Alpin.)

Training

The training was done for 2 epochs. I used 2 x RTX 6000s GPUs graciously provided by Kubernetes Bad for the full-parameter fine-tuning of the model.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	23.90
IFEval (0-Shot)	47.82
BBH (3-Shot)	32.54
MATH Lvl 5 (4-Shot)	12.61
GPQA (0-shot)	6.94
MuSR (0-shot)	14.01
MMLU-PRO (5-shot)	29.49