|
--- |
|
base_model: bobofrut/ladybird-base-7B-v8 |
|
license: apache-2.0 |
|
language: |
|
- en |
|
pipeline_tag: text-generation |
|
tags: |
|
- mistral |
|
- gguf |
|
library_name: llama.cpp |
|
model_creator: bobofrut |
|
model_name: ladybird base 7B v8 |
|
model_type: mistral |
|
prompt_template: | |
|
<|im_start|>system |
|
{system_message}<|im_end|> |
|
<|im_start|>user |
|
{prompt}<|im_end|> |
|
<|im_start|>assistant |
|
quantized_by: mgonzs13 |
|
--- |
|
|
|
# ladybird-base-7B-v8 |
|
|
|
**Model creator:** [bobofrut](https://huggingface.co/bobofrut)<br> |
|
**Original model**: [Mistroll-7B-v2.2](https://huggingface.co/bobofrut/ladybird-base-7B-v8)<br> |
|
**GGUF quantization:** `llama.cpp` commit [b8c1476e44cc1f3a1811613f65251cf779067636](https://github.com/ggerganov/llama.cpp/tree/b8c1476e44cc1f3a1811613f65251cf779067636)<br> |
|
|
|
## Description |
|
|
|
Ladybird-base-7B-v8 is based on the Mistral architecture, which is known for its efficiency and effectiveness in handling complex language understanding and generation tasks. The model incorporates several innovative architecture choices to enhance its performance: |
|
|
|
- **Grouped-Query Attention**: Optimizes attention mechanisms by grouping queries, reducing computational complexity while maintaining model quality. |
|
- **Sliding-Window Attention**: Improves the model's ability to handle long-range dependencies by focusing on relevant segments of input, enhancing understanding and coherence. |
|
- **Byte-fallback BPE Tokenizer**: Offers robust tokenization by combining the effectiveness of Byte-Pair Encoding (BPE) with a fallback mechanism for out-of-vocabulary bytes, ensuring comprehensive language coverage. |
|
|
|
## Prompt Template |
|
|
|
The prompt template is ChatML. |
|
|
|
``` |
|
<|im_start|>system |
|
{system_message}<|im_end|> |
|
<|im_start|>user |
|
{prompt}<|im_end|> |
|
<|im_start|>assistant |
|
``` |
