INDEX

Explanations

phrases indicating a measure of progress or extent in a specific context

oai_token-act-pair · gpt-3.5-turbo

claims about events that have not been substantiated

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

 Suppose

-0.72

thora

-0.68

 PLUS

-0.64

ammy

-0.63

 wanna

-0.63

 jeopardy

-0.61

 temptation

-0.61

 CODE

-0.60

 Difference

-0.59

 Eternity

-0.58

POSITIVE LOGITS

 ignored

1.20

 never

1.12

 seldom

1.09

 spared

1.05

 none

1.04

 rarely

1.04

never

1.03

 avoided

1.03

 nowhere

1.01

 remained

1.01

Activations Density 0.834%

phrases indicating a measure of progress or extent in a specific context

claims about events that have not been substantiated

No Comments

No Known Activations