SeaWolf-AI/Darwin-Darwin-27B-Opus-x-QuettaLLMs-27B-Korea-08274

Darwin V6 Evolved Model

Created by Darwin V6 diagnostic-guided evolutionary merge engine.

Parent Models

• Father: FINAL-Bench/Darwin-27B-Opus
• Mother: NewenAI/QuettaLLMs-27B-Koreasoner-V3

Evolution Result

• Benchmark score: 0.8274
• Merge method: slerp
• Merge hash: b247b9c7

Merge Statistics

• Total tensors merged: 1199
• Transplant A (Father preserved): 0
• Transplant B (Mother preserved): 0
• Blended: 1199

Optimal Genome

  global_ratio: 0.4812
  attn_ratio: 0.0681
  ffn_ratio: 0.9334
  embed_ratio: 0.3678
  density_a: 0.9699
  density_b: 0.9767
  block_0_ratio: 0.6041
  block_1_ratio: 0.4107
  block_2_ratio: 0.3975
  block_3_ratio: 0.6078
  block_4_ratio: 0.7820
  block_5_ratio: 0.3960
  block_6_ratio: 0.5333
  block_7_ratio: 0.6535
  block_8_ratio: 0.6938
  block_9_ratio: 0.6635
  mri_trust: 0.7493
  merge_method_weight: 0.6116

MRI Prescription Summary

• Average ratio_b: 0.500
• Attention ratio: 0.500
• FFN ratio: 0.500
• Embed ratio: 0.500
• Transplant A: 0
• Transplant B: 0
• Blended: 1199

Health Check

failed: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/data/ray_temp/ginipick/darwin_merge_cache/merged_bb10657e'. Use repo_type argument if needed.

Method

Darwin V6 implements DARE-TIES merge directly via PyTorch tensor operations. Per-tensor ratios are determined by MRI diagnostic (static tensor analysis + probe-based functional importance) combined with evolutionary genome search.

Formula: final_ratio = mri_ratio * mri_trust + genome_ratio * (1 - mri_trust)

DARE-TIES algorithm: Yadav et al., 2023 (re-implemented, not library-dependent)

Built by VIDRAFT. Apache 2.0.