INDEX

Explanations

specific names or terms, such as "Ogden," "Guerre," and "ligands."

oai_token-act-pair · gpt-3.5-turbo

names of individuals and entities associated with notable achievements or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 orbiting

-0.72

 Haas

-0.71

pter

-0.69

iflower

-0.69

icter

-0.68

 McDonnell

-0.66

boarding

-0.65

jong

-0.64

 Nadu

-0.63

EMBER

-0.63

POSITIVE LOGITS

glers

0.93

gers

0.85

ghan

0.80

vernment

0.78

ration

0.76

uchi

0.76

elson

0.76

Og

0.74

uracy

0.73

rahim

0.71

Activations Density 0.030%

specific names or terms, such as "Ogden," "Guerre," and "ligands."

names of individuals and entities associated with notable achievements or events

No Comments

No Known Activations