INDEX

Explanations

specific locations within a setting, especially focusing on rooms and areas

oai_token-act-pair · gpt-3.5-turbo

locations or areas within a physical space

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

Fact

-0.76

ETF

-0.74

apter

-0.72

ever

-0.69

Reviewer

-0.68

Course

-0.65

olver

-0.64

ortunately

-0.64

vasive

-0.64

ancial

-0.63

POSITIVE LOGITS

 doorway

0.94

 chairs

0.92

antry

0.91

 aisle

0.90

 chair

0.86

 hallway

0.84

 toilets

0.83

 fireplace

0.83

 courtyard

0.82

 window

0.82

Activations Density 0.157%

specific locations within a setting, especially focusing on rooms and areas

locations or areas within a physical space

No Comments

No Known Activations