INDEX

Explanations

words related to reporting or confirming information

oai_token-act-pair · gpt-3.5-turbo

phrases indicating confirmation or reporting of events or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

istries

-0.64

ospel

-0.64

ierre

-0.62

raphics

-0.59

 advice

-0.58

ggles

-0.56

folios

-0.56

 impression

-0.56

 compulsion

-0.56

 gears

-0.55

POSITIVE LOGITS

by

0.88

ibly

0.78

 Parenthood

0.77

 elsewhere

0.75

 anecd

0.75

LY

0.72

 beforehand

0.70

 incidentally

0.67

 clinically

0.67

 anew

0.67

Activations Density 0.146%

words related to reporting or confirming information

phrases indicating confirmation or reporting of events or information

No Comments

No Known Activations