INDEX

Explanations

instances of words related to intense emotions or actions

oai_token-act-pair · gpt-3.5-turbo

names and identifying terms related to specific people or entities

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

rast

-0.82

Ģ

-0.79

į

-0.76

Frames

-0.75

ersed

-0.75

apon

-0.74

ADRA

-0.73

 Kuro

-0.73

apons

-0.73

ESH

-0.72

POSITIVE LOGITS

stown

0.74

 wills

0.74

mint

0.73

enture

0.70

ithing

0.69

 Yards

0.67

 Belt

0.65

idges

0.65

 tongues

0.65

inh

0.65

Activations Density 0.059%

instances of words related to intense emotions or actions

names and identifying terms related to specific people or entities

No Comments

No Known Activations