INDEX

Explanations

instances where actions or events are suggested to happen differently than the norm

oai_token-act-pair · gpt-3.5-turbo

phrases or constructions that specify alternatives or comparisons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

CSS

-0.72

may

-0.72

izens

-0.69

cia

-0.69

soon

-0.68

WIND

-0.67

cue

-0.67

Corn

-0.66

RAW

-0.64

Aren

-0.64

POSITIVE LOGITS

 necessarily

1.16

 relying

1.14

 bothering

1.10

 merely

1.03

 simply

0.93

 letting

0.93

 risking

0.90

 outright

0.90

 allowing

0.86

 focusing

0.85

Activations Density 0.060%

instances where actions or events are suggested to happen differently than the norm

phrases or constructions that specify alternatives or comparisons

No Comments

No Known Activations