INDEX

Explanations

words related to news reporting, possibly focusing on specific locations or events

oai_token-act-pair · gpt-3.5-turbo

references to ports and port-related terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ority

-0.89

orig

-0.72

ites

-0.69

oning

-0.69

itary

-0.68

uum

-0.66

omet

-0.65

esian

-0.64

 numbered

-0.63

was

-0.62

POSITIVE LOGITS

PORT

4.00

LEASE

1.40

udeau

1.04

VIEW

0.96

SAN

0.93

plement

0.92

pport

0.85

 Milo

0.83

SAN

0.82

coni

0.81

Activations Density 0.034%

words related to news reporting, possibly focusing on specific locations or events

references to ports and port-related terminology

No Comments

No Known Activations