INDEX

Explanations

references to specific geographic locations, especially those related to the ocean

oai_token-act-pair · gpt-3.5-turbo

references to the ocean

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

ngth

-0.79

ENCE

-0.75

TI

-0.75

oulder

-0.73

enez

-0.72

Reloaded

-0.71

TOP

-0.69

enance

-0.68

ERAL

-0.68

ching

-0.64

POSITIVE LOGITS

 Ocean

1.21

ographer

0.96

 Spray

0.87

cean

0.86

 Seas

0.85

 Atlantic

0.83

 Beach

0.83

Ocean

0.83

 Knot

0.83

odon

0.83

Activations Density 0.013%

references to specific geographic locations, especially those related to the ocean

references to the ocean

No Comments

No Known Activations