INDEX

Explanations

mentions of things being in specific forms or formats

oai_token-act-pair · gpt-3.5-turbo

specific phrases or concepts indicated by the word "form."

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

avorite

-0.72

 sear

-0.69

Ban

-0.67

 Thro

-0.65

dain

-0.65

EStreamFrame

-0.64

 Duty

-0.63

 Motorsport

-0.63

 Rare

-0.63

 Hots

-0.62

POSITIVE LOGITS

aldehyde

1.32

ative

1.04

ulating

0.95

fitting

0.93

ulator

0.86

atives

0.84

idable

0.81

ulates

0.81

acion

0.81

ÑĮ

0.78

Activations Density 0.015%

mentions of things being in specific forms or formats

specific phrases or concepts indicated by the word "form."

No Comments

No Known Activations