INDEX

Explanations

mentions of a specific entity, "Lab" with variations in text formatting

oai_token-act-pair · gpt-3.5-turbo

references to specific labs

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

 odds

-0.86

 Marketable

-0.76

 understatement

-0.71

theless

-0.70

FUL

-0.67

fulness

-0.66

DIT

-0.65

 straw

-0.64

 LAST

-0.63

ãģį

-0.62

POSITIVE LOGITS

yrinth

1.46

rador

1.23

orers

1.20

elled

1.11

els

1.11

eled

1.09

emark

0.94

eling

0.94

orer

0.94

VIEW

0.94

Activations Density 0.019%

mentions of a specific entity, "Lab" with variations in text formatting

references to specific labs

No Comments

No Known Activations