INDEX

Explanations

phrases related to official statements or reports containing recommendations or instructions

oai_token-act-pair · gpt-3.5-turbo

the word "following" indicating a list or sequence of items

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

abella

-0.78

Downloadha

-0.70

aus

-0.69

Ear

-0.68

cest

-0.67

obyl

-0.67

istence

-0.66

uca

-0.66

hma

-0.66

 Memories

-0.66

POSITIVE LOGITS

 sequence

0.77

 week

0.74

 excerpt

0.73

 sections

0.70

 weeks

0.68

:-

0.67

 subsections

0.67

 line

0.67

 section

0.66

 message

0.65

Activations Density 0.023%

phrases related to official statements or reports containing recommendations or instructions

the word "following" indicating a list or sequence of items

No Comments

No Known Activations