INDEX

Explanations

phrases related to different types of functions or actions

oai_token-act-pair · gpt-3.5-turbo

terms related to various forms of support, creation, and safety issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Reloaded

-0.83

SN

-0.69

bern

-0.68

ãĥ¤

-0.68

ARR

-0.67

ãĥĥãĥī

-0.67

Gar

-0.66

ERG

-0.65

AIR

-0.63

Ô

-0.63

POSITIVE LOGITS

etting

0.75

ateurs

0.74

 advis

0.71

 prevention

0.69

 indexes

0.66

 queens

0.64

 etiquette

0.62

 readiness

0.61

etter

0.60

 chops

0.60

Activations Density 0.814%

phrases related to different types of functions or actions

terms related to various forms of support, creation, and safety issues

No Comments

No Known Activations