INDEX

Explanations

questions starting with 'how'

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

•How

Degree

No Comments

Negative Logits

advertisement

-0.68

-0.67

iculture

-0.67

 Supplement

-0.66

UME

-0.66

ograph

-0.63

icipated

-0.61

izu

-0.59

UM

-0.58

inian

-0.58

POSITIVE LOGITS

 much

1.07

ls

0.97

beit

0.94

 prevalent

0.93

 messed

0.92

far

0.89

 resilient

0.89

 MUCH

0.86

 fragile

0.85

itzer

0.82

Activations Density 0.068%

questions starting with 'how'

No Comments

No Known Activations