INDEX

Explanations

sequences of symbols forming patterns

oai_token-act-pair · gpt-3.5-turbo

sequences of repetitive characters or patterns

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

 reins

-0.64

ãĥĬ

-0.63

 proactive

-0.63

 athe

-0.62

 interf

-0.61

helps

-0.61

gra

-0.61

FY

-0.58

fal

-0.58

onse

-0.57

POSITIVE LOGITS

amiya

0.80

 respectively

0.77

Train

0.72

izon

0.67

depending

0.64

ifice

0.63

robe

0.63

shire

0.62

ercise

0.61

train

0.61

Activations Density 0.346%

sequences of symbols forming patterns

sequences of repetitive characters or patterns

No Comments

No Known Activations