xellDart13/NebuIA-10.7B-DPO

merge

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

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

• Weyaxi/SauerkrautLM-UNA-SOLAR-Instruct
• kodonho/SolarM-SakuraSolar-SLERP

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: Weyaxi/SauerkrautLM-UNA-SOLAR-Instruct
        layer_range: [0, 48]
      - model: kodonho/SolarM-SakuraSolar-SLERP
        layer_range: [0, 48]
merge_method: slerp
base_model: Weyaxi/SauerkrautLM-UNA-SOLAR-Instruct
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
dtype: bfloat16

Model	AGIEval	GPT4All	TruthfulQA	Bigbench	Average
NebuIA-10.7B-DPO	48.38	74.87	72.57	45.74	60.39

AGIEval

Task	Version	Metric	Value		Stderr
agieval_aqua_rat	0	acc	27.56	±	2.81
		acc_norm	27.95	±	2.82
agieval_logiqa_en	0	acc	42.40	±	1.94
		acc_norm	42.86	±	1.94
agieval_lsat_ar	0	acc	27.39	±	2.95
		acc_norm	25.22	±	2.87
agieval_lsat_lr	0	acc	54.31	±	2.21
		acc_norm	55.10	±	2.20
agieval_lsat_rc	0	acc	69.89	±	2.80
		acc_norm	69.14	±	2.82
agieval_sat_en	0	acc	79.61	±	2.81
		acc_norm	80.10	±	2.79
agieval_sat_en_without_passage	0	acc	48.06	±	3.49
		acc_norm	47.57	±	3.49
agieval_sat_math	0	acc	42.73	±	3.34
		acc_norm	39.09	±	3.30

Average: 48.38%

GPT4All

Task	Version	Metric	Value		Stderr
arc_challenge	0	acc	60.67	±	1.43
		acc_norm	63.74	±	1.40
arc_easy	0	acc	83.08	±	0.77
		acc_norm	81.23	±	0.80
boolq	1	acc	88.44	±	0.56
hellaswag	0	acc	69.28	±	0.46
		acc_norm	86.71	±	0.34
openbookqa	0	acc	37.60	±	2.17
		acc_norm	48.00	±	2.24
piqa	0	acc	80.25	±	0.93
		acc_norm	80.20	±	0.93
winogrande	0	acc	75.77	±	1.20

Average: 74.87%

TruthfulQA

Task	Version	Metric	Value		Stderr
truthfulqa_mc	1	mc1	57.89	±	1.73
		mc2	72.57	±	1.49

Average: 72.57%

Bigbench

Task	Version	Metric	Value		Stderr
bigbench_causal_judgement	0	multiple_choice_grade	58.95	±	3.58
bigbench_date_understanding	0	multiple_choice_grade	63.41	±	2.51
bigbench_disambiguation_qa	0	multiple_choice_grade	37.60	±	3.02
bigbench_geometric_shapes	0	multiple_choice_grade	28.97	±	2.40
		exact_str_match	0.00	±	0.00
bigbench_logical_deduction_five_objects	0	multiple_choice_grade	28.20	±	2.01
bigbench_logical_deduction_seven_objects	0	multiple_choice_grade	21.86	±	1.56
bigbench_logical_deduction_three_objects	0	multiple_choice_grade	47.00	±	2.89
bigbench_movie_recommendation	0	multiple_choice_grade	44.00	±	2.22
bigbench_navigate	0	multiple_choice_grade	63.90	±	1.52
bigbench_reasoning_about_colored_objects	0	multiple_choice_grade	58.15	±	1.10
bigbench_ruin_names	0	multiple_choice_grade	41.96	±	2.33
bigbench_salient_translation_error_detection	0	multiple_choice_grade	38.48	±	1.54
bigbench_snarks	0	multiple_choice_grade	65.75	±	3.54
bigbench_sports_understanding	0	multiple_choice_grade	72.31	±	1.43
bigbench_temporal_sequences	0	multiple_choice_grade	63.10	±	1.53
bigbench_tracking_shuffled_objects_five_objects	0	multiple_choice_grade	24.64	±	1.22
bigbench_tracking_shuffled_objects_seven_objects	0	multiple_choice_grade	18.00	±	0.92
bigbench_tracking_shuffled_objects_three_objects	0	multiple_choice_grade	47.00	±	2.89

Average: 45.74%

Average score: 60.39%