INDEX

Explanations

words related to news headlines or current events

oai_token-act-pair · gpt-3.5-turbo

possessive forms indicating ownership or association

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

wright

-1.00

este

-0.90

VO

-0.77

fo

-0.74

robe

-0.74

ova

-0.73

usa

-0.72

then

-0.71

eno

-0.70

cro

-0.70

POSITIVE LOGITS

 edition

1.11

 episode

0.98

 announcement

0.95

 installment

0.94

 iteration

0.90

 arrival

0.88

 editions

0.85

 deadline

0.85

 unveiling

0.85

 festivities

0.84

Activations Density 0.042%

words related to news headlines or current events

possessive forms indicating ownership or association

No Comments

No Known Activations