INDEX

Explanations

phrases related to repeated occurrences or ongoing presence

oai_token-act-pair · gpt-3.5-turbo

recurring themes or concepts

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

 Siber

-0.80

imore

-0.77

smanship

-0.75

jri

-0.70

fax

-0.69

eering

-0.69

mson

-0.69

 Emirates

-0.68

nton

-0.67

zynski

-0.67

POSITIVE LOGITS

entimes

1.08

etheless

0.86

 perenn

0.81

Lost

0.69

 endlessly

0.68

conservancy

0.67

 incess

0.67

luster

0.67

famous

0.66

 perpetually

0.65

Activations Density 0.034%

phrases related to repeated occurrences or ongoing presence

recurring themes or concepts

No Comments

No Known Activations