|
--- |
|
tags: |
|
- llama |
|
- instruct |
|
- finetune |
|
- chatml |
|
- gpt4 |
|
- synthetic data |
|
- distillation |
|
model-index: |
|
- name: Meta-Llama-3.1-8B-openhermes-2.5 |
|
results: [] |
|
license: apache-2.0 |
|
language: |
|
- en |
|
library_name: transformers |
|
datasets: |
|
- teknium/OpenHermes-2.5 |
|
--- |
|
|
|
# Model Card for Meta-Llama-3.1-8B-openhermes-2.5 |
|
|
|
This model is a fine-tuned version of Meta-Llama-3.1-8B on the OpenHermes-2.5 dataset. |
|
|
|
## Model Details |
|
|
|
### Model Description |
|
|
|
This is a fine-tuned version of the Meta-Llama-3.1-8B model, trained on the OpenHermes-2.5 dataset. It is designed for instruction following and general language tasks. |
|
|
|
- **Developed by:** artificialguybr |
|
- **Model type:** Causal Language Model |
|
- **Language(s):** English |
|
- **License:** apache-2.0 |
|
- **Finetuned from model:** meta-llama/Meta-Llama-3.1-8B |
|
|
|
### Model Sources |
|
|
|
- **Repository:** https://huggingface.co/artificialguybr/Meta-Llama-3.1-8B-openhermes-2.5 |
|
|
|
## Uses |
|
|
|
This model can be used for various natural language processing tasks, particularly those involving instruction following and general language understanding. |
|
|
|
### Direct Use |
|
|
|
The model can be used for tasks such as text generation, question answering, and other language-related applications. |
|
|
|
### Out-of-Scope Use |
|
|
|
The model should not be used for generating harmful or biased content. Users should be aware of potential biases in the training data. |
|
|
|
## Training Details |
|
|
|
### Training Data |
|
|
|
The model was fine-tuned on the teknium/OpenHermes-2.5 dataset. |
|
|
|
### Training Procedure |
|
|
|
#### Training Hyperparameters |
|
|
|
- **Training regime:** BF16 mixed precision |
|
- **Optimizer:** AdamW |
|
- **Learning rate:** Started at 0.00000249316296439037 (decaying) |
|
- **Batch size:** Not specified (gradient accumulation steps: 8) |
|
- **Training steps:** 13,368 |
|
- **Evaluation strategy:** Steps (every 0.16666666666666666 steps) |
|
- **Gradient checkpointing:** Enabled |
|
- **Weight decay:** 0 |
|
|
|
#### Hardware and Software |
|
|
|
- **Hardware:** NVIDIA A100-SXM4-80GB (1 GPU) |
|
- **Software Framework:** 🤗 Transformers, Axolotl |
|
|
|
## Evaluation |
|
|
|
### Metrics |
|
|
|
- **Loss:** 0.6727465987205505 (evaluation) |
|
- **Perplexity:** Not provided |
|
|
|
### Results |
|
|
|
- **Evaluation runtime:** 2,676.4173 seconds |
|
- **Samples per second:** 18.711 |
|
- **Steps per second:** 18.711 |
|
|
|
## Model Architecture |
|
|
|
- **Model Type:** LlamaForCausalLM |
|
- **Hidden size:** 4,096 |
|
- **Intermediate size:** 14,336 |
|
- **Number of attention heads:** Not specified |
|
- **Number of layers:** Not specified |
|
- **Activation function:** SiLU |
|
- **Vocabulary size:** 128,256 |
|
|
|
## Limitations and Biases |
|
|
|
More information is needed about specific limitations and biases of this model. |
|
|