INDEX

Explanations

verbs indicating causation or influence

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate causality or consequences

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

quickShipAvailable

-0.83

Enh

-0.70

anooga

-0.70

 fortun

-0.65

FRI

-0.63

Apps

-0.61

FX

-0.60

soType

-0.60

tons

-0.59

 Bene

-0.59

POSITIVE LOGITS

 stumble

1.02

 pursue

0.90

 confront

0.89

 ponder

0.88

 take

0.88

 become

0.87

 seek

0.87

 reach

0.86

iling

0.85

 conclude

0.85

Activations Density 0.143%

verbs indicating causation or influence

phrases that indicate causality or consequences

No Comments

No Known Activations