INDEX

Explanations

interjections expressing surprise or disbelief

oai_token-act-pair · gpt-3.5-turbo

expressions of surprise or realizations

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

•layer 5 - surprise

No Comments

Negative Logits

IUM

-0.82

 Construct

-0.70

-+-+

-0.70

*/(

-0.70

IAL

-0.68

eers

-0.68

ãĤ¼ãĤ¦ãĤ¹

-0.66

 fullest

-0.65

":[{"

-0.64

 Awakens

-0.63

POSITIVE LOGITS

hhhh

0.95

hh

0.93

oho

0.92

umph

0.90

oh

0.90

anian

0.86

warts

0.84

hhh

0.83

oy

0.82

awk

0.81

Activations Density 0.012%

interjections expressing surprise or disbelief

expressions of surprise or realizations

No Comments

No Known Activations