---
library_name: peft
tags:
- generated_from_trainer
base_model: llama-lang-adapt/pretrain-wura
model-index:
- name: africa-it-lora
  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/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.0`
```yaml
base_model: llama-lang-adapt/pretrain-wura
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true

load_in_8bit: true
load_in_4bit: false
strict: false

datasets:
  - path: llama-lang-adapt/african-it
    type: alpaca
    train_on_split: train
dataset_prepared_path: data/prepared-african-it

test_datasets:
  - path: llama-lang-adapt/african-it
    type: alpaca
    split: validation
    
output_dir: ./lora-out

sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00002

train_on_inputs: false
group_by_length: false
bf16: auto
fp16: false
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: 100
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
weight_decay: 0.01
fsdp:
fsdp_config:
special_tokens:

```

</details><br>

# african-it-lora

This model is a fine-tuned version of [llama-lang-adapt/pretrain-wura](https://huggingface.co/llama-lang-adapt/pretrain-wura) on the [llama-lang-adapt/african-it](https://huggingface.co/llama-lang-adapt/african-it) dataset.
It achieves the following result on the evaluation portion of that set:
- Loss: 0.5325

## 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: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 8
- total_train_batch_size: 16
- total_eval_batch_size: 2
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 1

### Training results

| Training Loss | Epoch | Step  | Validation Loss |
|:-------------:|:-----:|:-----:|:---------------:|
| 1.9958        | 0.0   | 1     | 3.1722          |
| 1.2509        | 0.25  | 7822  | 0.5396          |
| 1.0996        | 0.5   | 15644 | 0.5335          |
| 1.0109        | 0.75  | 23466 | 0.5321          |
| 1.0528        | 1.0   | 31288 | 0.5325          |


### Framework versions

- PEFT 0.9.0
- Transformers 4.40.0.dev0
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.0