INDEX

Explanations

numbers that indicate a reduction or decrease

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a reduction or simplification

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

oris

-0.76

awn

-0.73

went

-0.73

mir

-0.71

ruly

-0.69

ear

-0.67

ris

-0.67

 echoed

-0.66

rises

-0.65

hur

-0.64

POSITIVE LOGITS

 ashes

0.98

 scraps

0.94

 rubble

0.93

 zero

0.90

 fractions

0.88

 basics

0.88

 earth

0.87

 obscurity

0.87

 shorts

0.87

 mere

0.86

Activations Density 0.162%

numbers that indicate a reduction or decrease

phrases indicating a reduction or simplification

No Comments

No Known Activations