INDEX

Explanations

words related to surprising or impressing actions or events

oai_token-act-pair · gpt-3.5-turbo

expressions related to feelings of surprise or impressiveness

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

glas

-0.76

 concentrated

-0.65

sun

-0.65

raltar

-0.62

inatory

-0.62

 denying

-0.61

 uneven

-0.61

porary

-0.58

 misplaced

-0.56

 basis

-0.55

POSITIVE LOGITS

ingly

0.97

brate

0.85

brates

0.83

us

0.80

 herself

0.79

 yourselves

0.78

 himself

0.76

him

0.75

 ourselves

0.75

ively

0.74

Activations Density 0.400%

words related to surprising or impressing actions or events

expressions related to feelings of surprise or impressiveness

No Comments

No Known Activations