Qwen3-8B-YOYO-slerp-128K/README.md

license, language, base_model, pipeline_tag, tags

license	language	base_model	pipeline_tag	tags
apache-2.0	en zh	deepseek-ai/DeepSeek-R1-0528-Qwen3-8B Qwen/Qwen3-8B	text-generation	merge

Model Highlights:

• merge method: slerp

• Highest precision: dtype: float32 + out_dtype: bfloat16

• Brand-new chat template: ensures normal operation on LM Studio

• Context length: 131072

Model Selection Table:

Model	Context	Uses Basic Model
Qwen3-8B-YOYO-slerp	32K	Yes
Qwen3-8B-YOYO-slerp-128K	128K	Yes
Qwen3-8B-YOYO-nuslerp	32K	No
Qwen3-8B-YOYO-nuslerp-128K	128K	No
Qwen3-8B-YOYO-nuslerp-plus	32K	Yes
Qwen3-8B-YOYO-nuslerp-plus-128K	128K	Yes

Warning

: Models with 128K context may have slight quality loss. In most cases, please use the 32K native context!

Parameter Settings:

Thinking Mode:

Note

Temperature=0.6, TopP=0.95, TopK=20,MinP=0.

Configuration:

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: deepseek-ai/DeepSeek-R1-0528-Qwen3-8B
        layer_range: [0, 36]
      - model: Qwen/Qwen3-8B
        layer_range: [0, 36]
merge_method: slerp
base_model: Qwen/Qwen3-8B
parameters:
  t:
    - filter: self_attn
      value: [0, 0.5, 0.3, 0.7, 1]
    - filter: mlp
      value: [1, 0.5, 0.7, 0.3, 0]
    - value: 0.5
tokenizer_source: base
dtype: float32
out_dtype: bfloat16