INDEX

Explanations

it seems to be detecting specific phrases or keywords, but based solely on the given activations provided, it's not clear what the neuron is specifically looking for

oai_token-act-pair · gpt-3.5-turbo

mentions of economic concepts or factors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 subpoena

-0.86

 redacted

-0.79

 Hayden

-0.79

 warrant

-0.77

 Adams

-0.76

 Rutherford

-0.76

 subpoen

-0.75

 Belichick

-0.74

 Clapper

-0.73

 Cheney

-0.73

POSITIVE LOGITS

Vill

1.31

Residents

1.26

 Rohing

1.16

 villagers

1.16

Tour

1.12

Girls

1.10

Farm

1.07

Women

1.07

obyl

1.06

India

1.05

Activations Density 0.589%

it seems to be detecting specific phrases or keywords, but based solely on the given activations provided, it's not clear what the neuron is specifically looking for

mentions of economic concepts or factors

No Comments

No Known Activations