README.md

---
base_model: NousResearch/Meta-Llama-3-8B
tags:
- generated_from_trainer
model-index:
- name: llama3-8b-redmond-code290k
  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: NousResearch/Meta-Llama-3-8B
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: b-mc2/sql-create-context
    type: context_qa.load_v2
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./artificialguybr/llama3-8b-redmond-code290k

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

wandb_project: artificialguybr/llama3-8b-redmond-code290k
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 3
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
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 100
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

```

</details><br>

# LLAMA 3 8B Redmond CODE 290K

Thanks to [Redmond.ai](https://redmond.ai) for the GPU Support!

This model is a fine-tuned version of [NousResearch/Meta-Llama-3-8B](https://huggingface.co/NousResearch/Meta-Llama-3-8B) on the [ajibawa-2023/Code-290k-ShareGPT](https://huggingface.co/datasets/ajibawa-2023/Code-290k-ShareGPT) dataset.

## Model description

The Code-290k-ShareGPT model is a large language model designed to generate code and explanations in various programming languages, including Python, Java, JavaScript, GO, C++, Rust, Ruby, SQL, MySQL, R, Julia, Haskell, and more. It takes as input a prompt or question and outputs a corresponding code snippet with a detailed explanation.

The model is trained on a massive dataset of approximately 290,000 conversations, each consisting of two conversations. This dataset is in the Vicuna/ShareGPT format, which allows for efficient training and fine-tuning of the model.

The model is intended to be used in applications where code generation and explanation are necessary, such as coding assistance, education, and knowledge sharing.

## Intended uses & limitations
Intended uses:

Generating code and explanations in various programming languages

Assisting in coding tasks and education

Providing knowledge sharing and documentation

Integrating with other language models or tools to provide a more comprehensive coding experience

Limitations:

The model may not perform well on very rare or niche programming languages

The model may not generalize well to unseen coding styles or conventions

The model may not be able to handle extremely complex code or edge cases

The model may not be able to provide explanations for highly abstract or theoretical concepts

The model may not be able to handle ambiguous or open-ended prompts## 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
- gradient_accumulation_steps: 8
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 100
- num_epochs: 2

### Training results

Soon

### Framework versions

- Transformers 4.40.0.dev0
- Pytorch 2.2.2+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0