INDEX

Explanations

words indicating logical conclusions or implications

oai_token-act-pair · gpt-3.5-turbo

instances of the word "therefore."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.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

Not in Any Lists

No Comments

Negative Logits

 Columb

-0.75

ten

-0.71

 Feld

-0.70

 Straw

-0.67

 Blaz

-0.66

 Ashton

-0.64

 Lancaster

-0.63

jar

-0.62

 Wooden

-0.62

 Ventura

-0.61

POSITIVE LOGITS

ĵĺ

0.97

 guiActiveUn

0.93

efficients

0.92

 therefore

0.91

rehend

0.89

ratulations

0.84

 unfocusedRange

0.84

ħĭ

0.84

 occas

0.83

Ļ½

0.82

Activations Density 0.008%

words indicating logical conclusions or implications

instances of the word "therefore."

No Comments

No Known Activations