meta-llama/Meta-Llama-3-8B-Instruct
meta-llama/Meta-Llama-3-8B library_name: transformers tags:
mergekit
merge license: other license_name: llama3 license_link: LICENSE extra_gated_prompt: >-
META LLAMA 3 COMMUNITY LICENSE AGREEMENT

Llama-3-DARE-v1-8B

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 meta-llama/Meta-Llama-3-8B as a base.

Models Merged

The following models were included in the merge:

meta-llama/Meta-Llama-3-8B-Instruct

Configuration

The following YAML configuration was used to produce this model:

yaml

base_model:
  model:
    path: meta-llama/Meta-Llama-3-8B
dtype: bfloat16
merge_method: dare_ties
parameters:
  int8_mask: 1.0
  normalize: 0.0
slices:
- sources:
  - layer_range: [0, 32]
    model:
      model:
        path: meta-llama/Meta-Llama-3-8B-Instruct
    parameters:
      density:
      - filter: embed_token
        value: 0.07928115257656705
      - filter: model.norm
        value: 0.8644655087483669
      - filter: lm_head
        value: 0.8644655087483669
      - filter: layers.0.
        value: 0.7023215975496551
      - filter: layers.1.
        value: 0.5930151025883963
      - filter: layers.2.
        value: 0.9235600142754441
      - filter: layers.3.
        value: 0.30002273183016587
      - filter: layers.4.
        value: 0.8718725802605614
      - filter: layers.5.
        value: 0.8431819361408258
      - filter: layers.6.
        value: 0.26586459452250655
      - filter: layers.7.
        value: 0.20565052604836487
      - filter: layers.8.
        value: 0.7865609337773642
      - filter: layers.9.
        value: 0.8707795208055977
      - filter: layers.10.
        value: 0.19128908483148235
      - filter: layers.11.
        value: 0.925076317937204
      - filter: layers.12.
        value: 0.3764120891817022
      - filter: layers.13.
        value: 0.1537759874314265
      - filter: layers.14.
        value: 0.8827103791905803
      - filter: layers.15.
        value: 0.12994574083319627
      - filter: layers.16.
        value: 0.9421382084146839
      - filter: layers.17.
        value: 0.19784189207350633
      - filter: layers.18.
        value: 0.9174803701757963
      - filter: layers.19.
        value: 0.018511680579063827
      - filter: layers.20.
        value: 0.6089701838828931
      - filter: layers.21.
        value: 0.10667940525089978
      - filter: layers.22.
        value: 0.7569387463687963
      - filter: layers.23.
        value: 0.8617573261932969
      - filter: layers.24.
        value: 0.8851543429501336
      - filter: layers.25.
        value: 0.788944964233173
      - filter: layers.26.
        value: 0.022117233408225547
      - filter: layers.27.
        value: 0.9534983183659907
      - filter: layers.28.
        value: 0.792604065605754
      - filter: layers.29.
        value: 0.861671120755631
      - filter: layers.30.
        value: 0.2948434371778307
      - filter: layers.31.
        value: 0.18891036670059388
      weight:
      - filter: embed_token
        value: 0.244303585022914
      - filter: model.norm.
        value: 0.7921741064842713
      - filter: lm_head
        value: 0.7921741064842713
      - filter: layers.0.
        value: 0.7781818034133037
      - filter: layers.1.
        value: 0.8847835098465024
      - filter: layers.2.
        value: 0.6397880690508684
      - filter: layers.3.
        value: 0.7937466750856308
      - filter: layers.4.
        value: 0.8172928460203015
      - filter: layers.5.
        value: 0.8653872213596285
      - filter: layers.6.
        value: 0.9693001060347038
      - filter: layers.7.
        value: 0.8929433929262818
      - filter: layers.8.
        value: 0.8148535632999063
      - filter: layers.9.
        value: 0.9540888550674099
      - filter: layers.10.
        value: 0.9588139020445848
      - filter: layers.11.
        value: 0.9210433265234956
      - filter: layers.12.
        value: 0.7406799222922607
      - filter: layers.13.
        value: 0.6955899334998573
      - filter: layers.14.
        value: 0.15982240434953007
      - filter: layers.15.
        value: 0.911878214829457
      - filter: layers.16.
        value: 0.03946607654759535
      - filter: layers.17.
        value: 0.8322149510147696
      - filter: layers.18.
        value: 0.3733540416200388
      - filter: layers.19.
        value: 0.6932851946473836
      - filter: layers.20.
        value: 0.8918513138412437
      - filter: layers.21.
        value: 0.08682354503629197
      - filter: layers.22.
        value: 0.017627513785178843
      - filter: layers.23.
        value: 0.5246168031857519
      - filter: layers.24.
        value: 0.9561735656673052
      - filter: layers.25.
        value: 0.9036966249098499
      - filter: layers.26.
        value: 0.3270824190389672
      - filter: layers.27.
        value: 0.8585505386205572
      - filter: layers.28.
        value: 0.768055374051272
      - filter: layers.29.
        value: 0.8343611910799547
      - filter: layers.30.
        value: 0.9794887267657514
      - filter: layers.31.
        value: 0.9143801295721963
  - layer_range: [0, 32]
    model:
      model:
        path: meta-llama/Meta-Llama-3-8B
    parameters:
      density:
      - filter: embed_token
        value: 0.8288618306009948
      - filter: model.norm
        value: 0.23292234562975161
      - filter: lm_head
        value: 0.23292234562975161
      - filter: layers.0.
        value: 0.372454056194375
      - filter: layers.1.
        value: 0.14253218919354635
      - filter: layers.2.
        value: 0.3078424546105084
      - filter: layers.3.
        value: 0.25930581838885686
      - filter: layers.4.
        value: 0.978321706776098
      - filter: layers.5.
        value: 0.05609446797559111
      - filter: layers.6.
        value: 0.24314602951396613
      - filter: layers.7.
        value: 0.911885464400789
      - filter: layers.8.
        value: 0.7414851452758723
      - filter: layers.9.
        value: 0.18093642039158353
      - filter: layers.10.
        value: 0.9084501788509814
      - filter: layers.11.
        value: 0.2484987778735853
      - filter: layers.12.
        value: 0.02864842329393736
      - filter: layers.13.
        value: 0.035996105784139906
      - filter: layers.14.
        value: 0.022814159286914156
      - filter: layers.15.
        value: 0.8982289455883636
      - filter: layers.16.
        value: 0.6749279016774873
      - filter: layers.17.
        value: 0.2231666521672218
      - filter: layers.18.
        value: 0.08042893992871515
      - filter: layers.19.
        value: 0.9245191619395147
      - filter: layers.20.
        value: 0.10245727381170405
      - filter: layers.21.
        value: 0.6351565505035355
      - filter: layers.22.
        value: 0.13631302757594085
      - filter: layers.23.
        value: 0.02093704907904486
      - filter: layers.24.
        value: 0.046387358073609594
      - filter: layers.25.
        value: 0.39078092530283026
      - filter: layers.26.
        value: 0.6422945171240722
      - filter: layers.27.
        value: 0.9356630912083387
      - filter: layers.28.
        value: 0.20795911007902343
      - filter: layers.29.
        value: 0.10607547240686495
      - filter: layers.30.
        value: 0.0142610579064266
      - filter: layers.31.
        value: 0.9041406411956623
      weight:
      - filter: embed_token
        value: 0.11199251330860797
      - filter: model.norm.
        value: 0.14400396011333005
      - filter: lm_head
        value: 0.14400396011333005
      - filter: layers.0.
        value: 0.04471455188049395
      - filter: layers.1.
        value: 0.9523895147273159
      - filter: layers.2.
        value: 0.8434787249757318
      - filter: layers.3.
        value: 0.8016501716390354
      - filter: layers.4.
        value: 0.19555927362126765
      - filter: layers.5.
        value: 0.11657419158719552
      - filter: layers.6.
        value: 0.08984925597011956
      - filter: layers.7.
        value: 0.2629227995683227
      - filter: layers.8.
        value: 0.9596221742529512
      - filter: layers.9.
        value: 0.754944626119186
      - filter: layers.10.
        value: 0.6975471541191236
      - filter: layers.11.
        value: 0.7057432892778266
      - filter: layers.12.
        value: 0.8718457125948287
      - filter: layers.13.
        value: 0.08609329339067016
      - filter: layers.14.
        value: 0.062238388979337526
      - filter: layers.15.
        value: 0.06582279783729371
      - filter: layers.16.
        value: 0.16911181845635778
      - filter: layers.17.
        value: 0.2734099561761782
      - filter: layers.18.
        value: 0.21849327205885294
      - filter: layers.19.
        value: 0.128433065340601
      - filter: layers.20.
        value: 0.6468521489194113
      - filter: layers.21.
        value: 0.16152478595976968
      - filter: layers.22.
        value: 0.18505624743492918
      - filter: layers.23.
        value: 0.5230576418780342
      - filter: layers.24.
        value: 0.6970657976288235
      - filter: layers.25.
        value: 0.9119218240518024
      - filter: layers.26.
        value: 0.250155176705362
      - filter: layers.27.
        value: 0.9157060215586246
      - filter: layers.28.
        value: 0.9021879089954816
      - filter: layers.29.
        value: 0.1541974948279516
      - filter: layers.30.
        value: 0.8388637604851198
      - filter: layers.31.
        value: 0.06372200149189318

RichardErkhov/rmihaylov_-_Llama-3-DARE-v1-8B-8bits

Original model description:

META LLAMA 3 COMMUNITY LICENSE AGREEMENT

Llama-3-DARE-v1-8B

Merge Details

Merge Method

Models Merged

Configuration

RichardErkhov/
rmihaylov_-_Llama-3-DARE-v1-8B-8bits