INDEX

Explanations

phrases where someone is correcting or highlighting a particular point

oai_token-act-pair · gpt-3.5-turbo

references to making observations or highlighting points

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

DAQ

-0.92

uthor

-0.83

ccording

-0.78

perty

-0.76

eatures

-0.76

 notor

-0.75

 delinqu

-0.74

eco

-0.74

edom

-0.73

terness

-0.73

POSITIVE LOGITS

lessly

0.98

 pointing

0.87

point

0.82

points

0.79

 pointed

0.78

pointers

0.74

 forward

0.73

pointer

0.70

lessness

0.70

 blank

0.69

Activations Density 0.028%

phrases where someone is correcting or highlighting a particular point

references to making observations or highlighting points

No Comments

No Known Activations