56 lines
1.5 KiB
Markdown
56 lines
1.5 KiB
Markdown
|
---
|
||
|
|
base_model: vectionlabs/Maestro1-9B
|
||
|
|
library_name: transformers
|
||
|
|
license: apache-2.0
|
||
|
|
haffner:
|
||
|
|
readme_rev: 1
|
||
|
|
quantized_by: haffner
|
||
|
|
tags:
|
||
|
|
- heretic
|
||
|
|
- multimodal
|
||
|
|
- vision-language
|
||
|
|
- reasoning
|
||
|
|
- thinking
|
||
|
|
- math
|
||
|
|
- code
|
||
|
|
- long-context
|
||
|
|
- gguf
|
||
|
|
language:
|
||
|
|
- en
|
||
|
|
- it
|
||
|
|
---
|
||
|
|
## About
|
||
|
|
|
||
|
|
<p align="center">
|
||
|
|
<img src="MaestroBannerHeretic.jpg" width="100%">
|
||
|
|
</p>
|
||
|
|
|
||
|
|
<!-- ### quantize_version: 2 -->
|
||
|
|
<!-- ### output_tensor_quantised: 1 -->
|
||
|
|
<!-- ### convert_type: hf -->
|
||
|
|
<!-- ### tags: -->
|
||
|
|
<!-- ### quants: BF16 Q8\_0 Q6\_K Q5\_1 Q5\_K\_M Q5\_K\_S Q5\_0 Q4\_1 IQ4\_NL IQ4\_XS IQ3\_M IQ3\_S Q3\_K\_S Q3\_K\_M Q3\_K\_L Q2\_K TQ2\_0 TQ1\_0 Q1\_0
|
||
|
|
<!-- ### quants\_skip: -->
|
||
|
|
<!-- ### skip\_mmproj: -->
|
||
|
|
Heretic'ed safetensors and GGUF quants of https://huggingface.co/vectionlabs/Maestro1-9B
|
||
|
|
|
||
|
|
Multimodal Projector:
|
||
|
|
- [MMPROJ-BF16 - 1.16GB](https://huggingface.co/haffner/Maestro1-9B-Heretic/resolve/main/Maestro1-9B-heretic-mmproj-BF16.gguf?download=true)
|
||
|
|
- [MMPROJ-Q8_0 - 752MB](https://huggingface.co/haffner/Maestro1-9B-Heretic/resolve/main/Maestro1-9B-heretic-mmproj-Q8_0.gguf?download=true)
|
||
|
|
|
||
|
|
<!-- provided-files -->
|
||
|
|
|
||
|
|
Refusals: 3/100, KL divergence: 0.0741
|
||
|
|
|
||
|
|
* Parameters:
|
||
|
|
* direction_index = 18.77
|
||
|
|
* attn.o_proj.max_weight = 1.37
|
||
|
|
* attn.o_proj.max_weight_position = 27.86
|
||
|
|
* attn.o_proj.min_weight = 1.33
|
||
|
|
* attn.o_proj.min_weight_distance = 18.73
|
||
|
|
* mlp.down_proj.max_weight = 1.34
|
||
|
|
* mlp.down_proj.max_weight_position = 22.02
|
||
|
|
* mlp.down_proj.min_weight = 1.24
|
||
|
|
* mlp.down_proj.min_weight_distance = 14.79
|
||
|
|
|
||
|
|
<!-- end -->
|