INDEX

Explanations

proper nouns, particularly focusing on names

oai_token-act-pair · gpt-3.5-turbo

the name "Richards" in various contexts

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

phis

-1.01

phas

-0.91

anooga

-0.86

unity

-0.74

unal

-0.72

emort

-0.72

ivated

-0.72

cue

-0.71

ocular

-0.69

efully

-0.68

POSITIVE LOGITS

 Richards

0.88

cream

0.75

holder

0.73

zman

0.72

iders

0.72

hips

0.71

buster

0.70

DF

0.68

burgh

0.67

Hof

0.66

Activations Density 0.029%

proper nouns, particularly focusing on names

the name "Richards" in various contexts

No Comments

No Known Activations