INDEX

Explanations

superlatives and rankings related to different entities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating superlative attributes or rankings in the world

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

ints

-0.74

xual

-0.72

adel

-0.69

edin

-0.69

onomy

-0.68

ULL

-0.68

irement

-0.67

²¾

-0.66

IRED

-0.65

ikawa

-0.64

POSITIVE LOGITS

 largest

1.24

 tallest

1.16

 smallest

1.05

 foremost

1.05

 richest

1.04

 busiest

1.03

 fastest

1.02

 longest

1.00

 oldest

1.00

 hottest

1.00

Activations Density 0.087%

superlatives and rankings related to different entities

phrases indicating superlative attributes or rankings in the world

No Comments

No Known Activations