---
library_name: peft
base_model: hardlyworking/Noodles-Merge-12B
tags:
- axolotl
- generated_from_trainer
datasets:
- Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
- ResplendentAI/bluemoon
- hardlyworking/openerotica-freedomrp-sharegpt-system
- MinervaAI/Aesir-Preview
- anthracite-core/c2_logs_32k_v1.1
- Nitral-AI/Creative_Writing-ShareGPT
- PJMixers/lodrick-the-lafted_OpusStories-Story2Prompt-ShareGPT
model-index:
- name: Beef-12B
  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://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: `0.8.0`
```yaml
## model
base_model: hardlyworking/Noodles-Merge-12B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
## upload
hub_model_id: hardlyworking/Beef-12B
hub_strategy: "all_checkpoints"
push_dataset_to_hub:
hf_use_auth_token: true
## qlora COPE
load_in_8bit: false
load_in_4bit: false
strict: false

## data 
datasets:
  - path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
    type: dan-chat-advanced
  - path: ResplendentAI/bluemoon
    type: dan-chat-advanced
  - path: hardlyworking/openerotica-freedomrp-sharegpt-system
    type: dan-chat-advanced
  - path: MinervaAI/Aesir-Preview
    type: dan-chat-advanced
  - path: anthracite-core/c2_logs_32k_v1.1
    type: dan-chat-advanced
  - path: Nitral-AI/Creative_Writing-ShareGPT
    type: dan-chat-advanced
  - path: PJMixers/lodrick-the-lafted_OpusStories-Story2Prompt-ShareGPT
    type: dan-chat-advanced

shuffle_merged_datasets: true
dataset_prepared_path: dataset_prepared
val_set_size: 0.01
output_dir: outputs/out

## LIGER & CCE
plugins:
  - axolotl.integrations.liger.LigerPlugin
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
liger_rope: true
liger_rms_norm: true
liger_layer_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: false
cut_cross_entropy: false

## CTX settings
sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

## Lora 
adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
peft_use_rslora: true
lora_modules_to_save:
  - embed_tokens
  - lm_head

## WandB
wandb_project: JoeyBoy
wandb_entity:
wandb_watch:
wandb_name: 
wandb_log_model:

## evals
evals_per_epoch: 8
eval_table_size:
eval_max_new_tokens: 128

## hoe params
gradient_accumulation_steps: 2
micro_batch_size: 4
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

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
s2_attention:

warmup_steps: 40
saves_per_epoch: 2
debug:
## for ademiamix 
deepspeed: ./deepspeed_configs/zero3_bf16.json
## for adamw
## deepspeed: ./deepspeed_configs/zero3_bf16.json
weight_decay: 0.01
fsdp:
fsdp_config:
special_tokens:
   pad_token: <pad>

```

</details><br>

# Beef-12B

This model is a fine-tuned version of [hardlyworking/Noodles-Merge-12B](https://huggingface.co/hardlyworking/Noodles-Merge-12B) on the Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned, the ResplendentAI/bluemoon, the hardlyworking/openerotica-freedomrp-sharegpt-system, the MinervaAI/Aesir-Preview, the anthracite-core/c2_logs_32k_v1.1, the Nitral-AI/Creative_Writing-ShareGPT and the PJMixers/lodrick-the-lafted_OpusStories-Story2Prompt-ShareGPT datasets.
It achieves the following results on the evaluation set:
- Loss: 1.5655

## 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: 2e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- total_eval_batch_size: 16
- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT 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: 40
- num_epochs: 2.0

### Training results

| Training Loss | Epoch  | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 1.7653        | 0.0028 | 1    | 1.7865          |
| 1.3533        | 0.1255 | 45   | 1.6828          |
| 1.2807        | 0.2510 | 90   | 1.6545          |
| 1.3957        | 0.3766 | 135  | 1.6300          |
| 1.2727        | 0.5021 | 180  | 1.6176          |
| 1.2438        | 0.6276 | 225  | 1.6074          |
| 1.3147        | 0.7531 | 270  | 1.5958          |
| 1.2466        | 0.8787 | 315  | 1.5905          |
| 1.3144        | 1.0028 | 360  | 1.5844          |
| 1.1868        | 1.1283 | 405  | 1.5784          |
| 1.3102        | 1.2538 | 450  | 1.5750          |
| 1.2746        | 1.3794 | 495  | 1.5734          |
| 1.1794        | 1.5049 | 540  | 1.5692          |
| 1.2141        | 1.6304 | 585  | 1.5671          |
| 1.1795        | 1.7559 | 630  | 1.5660          |
| 1.4297        | 1.8815 | 675  | 1.5655          |


### Framework versions

- PEFT 0.15.1
- Transformers 4.51.3
- Pytorch 2.6.0+cu124
- Datasets 3.5.0
- Tokenizers 0.21.1