INDEX

Explanations

information related to statistics or likelihoods

oai_token-act-pair · gpt-3.5-turbo

references to likelihood or probability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

lain

-0.94

andan

-0.85

rief

-0.84

uart

-0.83

zeb

-0.83

eni

-0.80

artney

-0.80

aan

-0.79

uay

-0.79

ometimes

-0.78

POSITIVE LOGITS

 culprit

0.74

ties

0.73

 scenario

0.72

 assumption

0.72

 ancestor

0.71

 future

0.70

 linem

0.69

 underest

0.66

 underestimate

0.65

 inev

0.64

Activations Density 0.030%

information related to statistics or likelihoods

references to likelihood or probability

No Comments

No Known Activations