matsuo-llm-advanced-phase-f5b

merged_model

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using Qwen/Qwen2.5-7B-Instruct as a base.

Models Merged

The following models were included in the merge:

• ./outputs/phase_d/merged_model
• ./outputs/phase_e2b/merged_model

Configuration

The following YAML configuration was used to produce this model:

# F5-b: DARE-TIES Low Density (Phase D 0.60 + E2-b 0.40, density=0.40)
#
# F4シリーズの教訓:
#   F4-a (density=0.5, D:0.6/E2b:0.4): DB 52.44%, ALF 56%(pre) → 52%(official)
#   F4-b (density=0.5, D:0.7/E2b:0.3): DB 52.99%, ALF 52%(pre)
#   F4-c (density=0.7, D:0.6/E2b:0.4): DB 54.02%, ALF 52%(pre)
#
# F5-b戦略: F4-aと同じ重み(D:0.6/E2b:0.4)だが密度を0.5→0.40に下げる
# F4-aとの差分: density 0.50 → 0.40 のみ（1変数変更の原則）
# 期待: ALFへの影響が減少し、F4-aのALF 56%(pre)を安定化

models:
  - model: ./outputs/phase_d/merged_model
    parameters:
      density: 0.40
      weight: 0.60
  - model: ./outputs/phase_e2b/merged_model
    parameters:
      density: 0.40
      weight: 0.40
merge_method: dare_ties
base_model: Qwen/Qwen2.5-7B-Instruct
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16