---
base_model: Lambent/proto-nova-eidolon-v2alpha0.3-14B
tags:
- generated_from_trainer
- not-for-all-audiences
model-index:
- name: dpoq
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

<img src="https://cdn.midjourney.com/2cf1309c-bcde-41e1-bd58-957feccb3ed8/0_1.jpeg"></img>

This version has been tuned from the fascinating arcee-ai/SuperNova-Medius as root model.

Censorship remains notable on this one, just including the Not For All Audiences tag due to dataset.

EQ-Bench is about 1 point lower than its ancestor, but fixed a syntax issue. May indicate a bit of expected intelligence loss.

Methodology: A bit of custom fine-tuning, with the plurality from the 'filtered' subset of argilla/ifeval-like-data experimentally trained
with 'input/output' roles rather than 'user/assistant'
(other instruction sampling stayed chatml-style, some continued pretraining added with a bias to older public domain styles);
ties merged at full saturation with the original over base Qwen, then this DPO.

[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.1`
```yaml
base_model: Lambent/proto-nova-eidolon-v2alpha0.3-14B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
trust_remote_code: true

save_safetensors: true

load_in_8bit: false
load_in_4bit: true
strict: false

rl: dpo
# total_num_tokens: 
datasets:
  - path: Lambent/ai-deconditioning-synthesized-dpo
    split: train
    type: chatml.prompt_pairs
  - path: jondurbin/gutenberg-dpo-v0.1
    split: train
    type: chatml.prompt_pairs
  - path: nbeerbower/gutenberg2-dpo
    split: train
    type: chatml.prompt_pairs
  - path: unalignment/toxic-dpo-v0.2
    split: train
    type: chatml.prompt_pairs
  - path: vicgalle/configurable-system-prompt-multitask
    split: train
    type: chatml.prompt_pairs

dataset_prepared_path: prepared-dpo
output_dir: ./dpoq
val_set_size: 0.01

seed: 1

sequence_len: 2048
sample_packing: false
eval_sample_packing: false
pad_to_sequence_len: false

adapter: qlora
lora_model_dir:
lora_r: 256
lora_alpha: 256
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
peft_use_dora: true

wandb_project: eidolon-qwen2.5-qlora-dpo
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 16
micro_batch_size: 2
num_epochs: 1
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 0.00001
#cosine_min_lr_ratio: 0.1
#cosine_constant_lr_ratio: 0.95

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

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

warmup_steps: 16
evals_per_epoch: 8
saves_per_epoch: 8
save_total_limit: 2
debug:
deepspeed:
weight_decay: 0.001
fsdp:
fsdp_config:

```

</details><br>

# dpoq

This model is a fine-tuned version of [Lambent/proto-nova-eidolon-v2alpha0.3-14B](https://huggingface.co/Lambent/proto-nova-eidolon-v2alpha0.3-14B) on the None 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: 1e-05
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 16
- training_steps: 124

### Training results



### Framework versions

- PEFT 0.13.2
- Transformers 4.45.2
- Pytorch 2.3.1+cu121
- Datasets 3.0.1
- Tokenizers 0.20.1