---
library_name: transformers
tags:
- mergekit
- merge
---
# iceblink-v3d

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

## Merge Details
### Merge Method

This model was merged using the [SLERP](https://en.wikipedia.org/wiki/Slerp) merge method.

### Models Merged

The following models were included in the merge:
* [zai-org/GLM-4.5-Air](https://huggingface.co/zai-org/GLM-4.5-Air)
* [ApocalypseParty/derestricted-iceblink](https://huggingface.co/ApocalypseParty/derestricted-iceblink)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
models:
  - model: ApocalypseParty/derestricted-iceblink
  - model: zai-org/GLM-4.5-Air
merge_method: slerp
base_model: zai-org/GLM-4.5-Air
parameters:
  t:
    - filter: embed_tokens
      value: 0.0
    - filter: layernorm
      value: 0.0
    - filter: e_score_correction
      value: 0.0
    - filter: layers.46.
      value: 0.0

    - filter: mlp.gate.weight
      value: 0.25

    - filter: layers.0.
      value: 0.30

    - filter: mlp.shared
      value: 0.60

    - filter: layers.1.self_attn
      value: 0.30
    - filter: layers.2.self_attn
      value: 0.30
    - filter: layers.3.self_attn
      value: 0.28
    - filter: layers.4.self_attn
      value: 0.28
    - filter: layers.5.self_attn
      value: 0.26
    - filter: layers.6.self_attn
      value: 0.26
    - filter: layers.7.self_attn
      value: 0.24
    - filter: layers.8.self_attn
      value: 0.24
    - filter: layers.9.self_attn
      value: 0.20
    - filter: layers.10.self_attn
      value: 0.18
    - filter: layers.11.self_attn
      value: 0.18
    - filter: layers.12.self_attn
      value: 0.18
    - filter: layers.13.self_attn
      value: 0.18
    - filter: layers.14.self_attn
      value: 0.18
    - filter: layers.15.self_attn
      value: 0.18
    - filter: layers.16.self_attn
      value: 0.18
    - filter: layers.17.self_attn
      value: 0.18
    - filter: layers.18.self_attn
      value: 0.18
    - filter: layers.19.self_attn
      value: 0.20
    - filter: layers.20.self_attn
      value: 0.20
    - filter: layers.21.self_attn
      value: 0.22
    - filter: layers.22.self_attn
      value: 0.24
    - filter: layers.23.self_attn
      value: 0.28
    - filter: layers.24.self_attn
      value: 0.32
    - filter: layers.25.self_attn
      value: 0.36
    - filter: layers.26.self_attn
      value: 0.40
    - filter: layers.27.self_attn
      value: 0.44
    - filter: layers.28.self_attn
      value: 0.48
    - filter: layers.29.self_attn
      value: 0.50
    - filter: layers.30.self_attn
      value: 0.52
    - filter: layers.31.self_attn
      value: 0.54
    - filter: layers.32.self_attn
      value: 0.55
    - filter: layers.33.self_attn
      value: 0.55
    - filter: layers.34.self_attn
      value: 0.55
    - filter: layers.35.self_attn
      value: 0.55
    - filter: layers.36.self_attn
      value: 0.58
    - filter: layers.37.self_attn
      value: 0.60
    - filter: layers.38.self_attn
      value: 0.62
    - filter: layers.39.self_attn
      value: 0.62
    - filter: layers.40.self_attn
      value: 0.60
    - filter: layers.41.self_attn
      value: 0.58
    - filter: layers.42.self_attn
      value: 0.55
    - filter: layers.43.self_attn
      value: 0.52
    - filter: layers.44.self_attn
      value: 0.48
    - filter: layers.45.self_attn
      value: 0.45
      
    - filter: layers.1.mlp.experts
      value: 0.60
    - filter: layers.2.mlp.experts
      value: 0.62
    - filter: layers.3.mlp.experts
      value: 0.64
    - filter: layers.4.mlp.experts
      value: 0.66
    - filter: layers.5.mlp.experts
      value: 0.68
    - filter: layers.6.mlp.experts
      value: 0.70
    - filter: layers.7.mlp.experts
      value: 0.72
    - filter: layers.8.mlp.experts
      value: 0.74
    - filter: layers.9.mlp.experts
      value: 0.76
    - filter: layers.10.mlp.experts
      value: 0.78
    - filter: layers.11.mlp.experts
      value: 0.80
    - filter: layers.12.mlp.experts
      value: 0.80
    - filter: layers.13.mlp.experts
      value: 0.80
    - filter: layers.14.mlp.experts
      value: 0.80
    - filter: layers.15.mlp.experts
      value: 0.82
    - filter: layers.16.mlp.experts
      value: 0.82
    - filter: layers.17.mlp.experts
      value: 0.82
    - filter: layers.18.mlp.experts
      value: 0.82
    - filter: layers.19.mlp.experts
      value: 0.82
    - filter: layers.20.mlp.experts
      value: 0.82
    - filter: layers.21.mlp.experts
      value: 0.82
    - filter: layers.22.mlp.experts
      value: 0.82
    - filter: layers.23.mlp.experts
      value: 0.82
    - filter: layers.24.mlp.experts
      value: 0.82
    - filter: layers.25.mlp.experts
      value: 0.82
    - filter: layers.26.mlp.experts
      value: 0.82
    - filter: layers.27.mlp.experts
      value: 0.82
    - filter: layers.28.mlp.experts
      value: 0.82
    - filter: layers.29.mlp.experts
      value: 0.82
    - filter: layers.30.mlp.experts
      value: 0.82
    - filter: layers.31.mlp.experts
      value: 0.85
    - filter: layers.32.mlp.experts
      value: 0.85
    - filter: layers.33.mlp.experts
      value: 0.85
    - filter: layers.34.mlp.experts
      value: 0.85
    - filter: layers.35.mlp.experts
      value: 0.85
    - filter: layers.36.mlp.experts
      value: 0.85
    - filter: layers.37.mlp.experts
      value: 0.85
    - filter: layers.38.mlp.experts
      value: 0.85
    - filter: layers.39.mlp.experts
      value: 0.85
    - filter: layers.40.mlp.experts
      value: 0.85
    - filter: layers.41.mlp.experts
      value: 0.82
    - filter: layers.42.mlp.experts
      value: 0.78
    - filter: layers.43.mlp.experts
      value: 0.75
    - filter: layers.44.mlp.experts
      value: 0.72
    - filter: layers.45.mlp.experts
      value: 0.70
    - value: 0.35
dtype: bfloat16
out_dtype: bfloat16
tokenizer:
  source: base

```