INDEX

Explanations

the word "Happ" or variations like "Happened" with varying degrees of activation strength

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Happ" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

trl

-0.91

ħĭ

-0.69

 caution

-0.68

 understanding

-0.66

 Nile

-0.66

 guiActiveUn

-0.65

HCR

-0.65

 IMAGES

-0.63

 explorers

-0.63

ĨĴ

-0.62

POSITIVE LOGITS

ening

1.05

ened

1.04

 Happ

0.94

olitics

0.91

earance

0.90

rox

0.87

iest

0.85

shake

0.83

ipeg

0.81

aday

0.81

Activations Density 0.004%

the word "Happ" or variations like "Happened" with varying degrees of activation strength

occurrences of the word "Happ" and its variations

No Comments

No Known Activations