INDEX

Explanations

numbers in addresses

oai_token-act-pair · gpt-3.5-turbo

numerical addresses and locations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ophon

-0.79

 regimes

-0.77

pse

-0.70

 gobl

-0.69

 unsc

-0.69

zbollah

-0.68

 endors

-0.68

otrop

-0.68

 subsystem

-0.67

 flavours

-0.67

POSITIVE LOGITS

Rue

1.16

 Pleasant

1.10

 Meadow

1.09

NW

1.07

 Avenue

1.06

 Woodward

1.06

 Riverside

1.06

 Broadway

1.04

 Sunset

1.02

 Huntington

1.01

Activations Density 0.039%

numbers in addresses

numerical addresses and locations

No Comments

No Known Activations