Edit model card

Llama-3-DARE-v2-8B

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:

Configuration

The following YAML configuration was used to produce this model:

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: 0.1312973174759711
      - filter: model.norm
        value: 0.9822573068643683
      - filter: lm_head
        value: 0.9822573068643683
      - filter: layers.0.
        value: 0.7585980531146381
      - filter: layers.1.
        value: 0.561884676687423
      - filter: layers.2.
        value: 0.9813332103529979
      - filter: layers.3.
        value: 0.21930890320643737
      - filter: layers.4.
        value: 0.8438595439416953
      - filter: layers.5.
        value: 0.7505829508553786
      - filter: layers.6.
        value: 0.25806652766624183
      - filter: layers.7.
        value: 0.09440963517101368
      - filter: layers.8.
        value: 0.7755855137764502
      - filter: layers.9.
        value: 0.980153628060705
      - filter: layers.10.
        value: 0.19005899180649255
      - filter: layers.11.
        value: 0.8607970083829785
      - filter: layers.12.
        value: 0.397155695730673
      - filter: layers.13.
        value: 0.13006504372382097
      - filter: layers.14.
        value: 0.09190339054787032
      - filter: layers.15.
        value: 0.14507208971010788
      - filter: layers.16.
        value: 0.8783567570968138
      - filter: layers.17.
        value: 0.12105889524706802
      - filter: layers.18.
        value: 0.8613122534393256
      - filter: layers.19.
        value: 0.04810799757747608
      - filter: layers.20.
        value: 0.5576780215726541
      - filter: layers.21.
        value: 0.027665563208104402
      - filter: layers.22.
        value: 0.7853866990645277
      - filter: layers.23.
        value: 0.8674754718436442
      - filter: layers.24.
        value: 0.915444005146838
      - filter: layers.25.
        value: 0.8408012482479134
      - filter: layers.26.
        value: 0.021388326989086454
      - filter: layers.27.
        value: 0.8382317625625855
      - filter: layers.28.
        value: 0.0418890647830244
      - filter: layers.29.
        value: 0.16347844552378538
      - filter: layers.30.
        value: 0.3152357116732385
      - filter: layers.31.
        value: 0.24560833074353305
      weight:
      - filter: embed_token
        value: 0.07729925635324639
      - filter: model.norm.
        value: 0.8178953444711945
      - filter: lm_head
        value: 0.8178953444711945
      - filter: layers.0.
        value: 0.7744866238001816
      - filter: layers.1.
        value: 0.9434402042495054
      - filter: layers.2.
        value: 0.7277347873270702
      - filter: layers.3.
        value: 0.7344542377948238
      - filter: layers.4.
        value: 0.7789550773026275
      - filter: layers.5.
        value: 0.9977070050825179
      - filter: layers.6.
        value: 0.9790331299458723
      - filter: layers.7.
        value: 0.936694814017198
      - filter: layers.8.
        value: 0.8022245662093622
      - filter: layers.9.
        value: 0.9207818987772527
      - filter: layers.10.
        value: 0.9177346250309115
      - filter: layers.11.
        value: 0.9705518504378707
      - filter: layers.12.
        value: 0.6962192295710468
      - filter: layers.13.
        value: 0.6601913931204128
      - filter: layers.14.
        value: 0.7700302171860424
      - filter: layers.15.
        value: 0.8764914331693985
      - filter: layers.16.
        value: 0.1680474521036102
      - filter: layers.17.
        value: 0.7252502118522168
      - filter: layers.18.
        value: 0.974742254778171
      - filter: layers.19.
        value: 0.6195931122778885
      - filter: layers.20.
        value: 0.8808789662609544
      - filter: layers.21.
        value: 0.10257515634421632
      - filter: layers.22.
        value: 0.04739908873803297
      - filter: layers.23.
        value: 0.9609916339993614
      - filter: layers.24.
        value: 0.919873783591387
      - filter: layers.25.
        value: 0.8920927224651306
      - filter: layers.26.
        value: 0.3163445951176943
      - filter: layers.27.
        value: 0.9183620719063076
      - filter: layers.28.
        value: 0.7696116230696886
      - filter: layers.29.
        value: 0.8092141535971384
      - filter: layers.30.
        value: 0.9831761728133218
      - filter: layers.31.
        value: 0.8368550180785097
  - layer_range: [0, 32]
    model:
      model:
        path: meta-llama/Meta-Llama-3-8B
    parameters:
      density:
      - filter: embed_token
        value: 0.17966590577034688
      - filter: model.norm
        value: 0.21103453748756532
      - filter: lm_head
        value: 0.21103453748756532
      - filter: layers.0.
        value: 0.1258111935267289
      - filter: layers.1.
        value: 0.14712518700082466
      - filter: layers.2.
        value: 0.15119918492525058
      - filter: layers.3.
        value: 0.2786737466113946
      - filter: layers.4.
        value: 0.7941421047966343
      - filter: layers.5.
        value: 0.14996187005944175
      - filter: layers.6.
        value: 0.21464495779281623
      - filter: layers.7.
        value: 0.9544388237818261
      - filter: layers.8.
        value: 0.6823542810137563
      - filter: layers.9.
        value: 0.12285240879708
      - filter: layers.10.
        value: 0.9560979622228943
      - filter: layers.11.
        value: 0.18774045206190226
      - filter: layers.12.
        value: 0.040667440606417166
      - filter: layers.13.
        value: 0.06577308115912563
      - filter: layers.14.
        value: 0.10898066770591856
      - filter: layers.15.
        value: 0.7988665902205875
      - filter: layers.16.
        value: 0.6902648842556923
      - filter: layers.17.
        value: 0.21929644905849668
      - filter: layers.18.
        value: 0.031818815520317414
      - filter: layers.19.
        value: 0.09373597067550557
      - filter: layers.20.
        value: 0.05663841107031176
      - filter: layers.21.
        value: 0.573566646668988
      - filter: layers.22.
        value: 0.2100277505470118
      - filter: layers.23.
        value: 0.14741793214499152
      - filter: layers.24.
        value: 0.13421247708500247
      - filter: layers.25.
        value: 0.3923811614341299
      - filter: layers.26.
        value: 0.7128080834275775
      - filter: layers.27.
        value: 0.7173419548629656
      - filter: layers.28.
        value: 0.25148835909796347
      - filter: layers.29.
        value: 0.01041679056502079
      - filter: layers.30.
        value: 0.04502964289907793
      - filter: layers.31.
        value: 0.7933910983298705
      weight:
      - filter: embed_token
        value: 0.1379551165348899
      - filter: model.norm.
        value: 0.1145495237372664
      - filter: lm_head
        value: 0.1145495237372664
      - filter: layers.0.
        value: 0.058088970620673676
      - filter: layers.1.
        value: 0.9812897914515345
      - filter: layers.2.
        value: 0.7981160133946364
      - filter: layers.3.
        value: 0.7634141926997863
      - filter: layers.4.
        value: 0.16065704042361334
      - filter: layers.5.
        value: 0.11985625529888592
      - filter: layers.6.
        value: 0.12337476562239155
      - filter: layers.7.
        value: 0.24780333206355964
      - filter: layers.8.
        value: 0.031049663247481193
      - filter: layers.9.
        value: 0.9399411230471199
      - filter: layers.10.
        value: 0.059148520666240975
      - filter: layers.11.
        value: 0.7231307420560733
      - filter: layers.12.
        value: 0.8604431238597543
      - filter: layers.13.
        value: 0.7874665970878372
      - filter: layers.14.
        value: 0.18061875009463604
      - filter: layers.15.
        value: 0.09518784175500823
      - filter: layers.16.
        value: 0.27798323894923355
      - filter: layers.17.
        value: 0.9413932817490506
      - filter: layers.18.
        value: 0.2512810899514295
      - filter: layers.19.
        value: 0.11709350712163674
      - filter: layers.20.
        value: 0.6739674929530162
      - filter: layers.21.
        value: 0.21208528089924455
      - filter: layers.22.
        value: 0.811536887881144
      - filter: layers.23.
        value: 0.6200821391879456
      - filter: layers.24.
        value: 0.7663192062414923
      - filter: layers.25.
        value: 0.8657672217068108
      - filter: layers.26.
        value: 0.10630047742154969
      - filter: layers.27.
        value: 0.9641883915469491
      - filter: layers.28.
        value: 0.9807033575417018
      - filter: layers.29.
        value: 0.04810198321663159
      - filter: layers.30.
        value: 0.8249574979326807
      - filter: layers.31.
        value: 0.01564957234872716
Downloads last month
20
Safetensors
Model size
8.03B params
Tensor type
BF16
Β·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for rmihaylov/Llama-3-DARE-v2-8B

Spaces using rmihaylov/Llama-3-DARE-v2-8B 4