INDEX

Explanations

the name "Luka"

oai_token-act-pair · gpt-3.5-turbo

repeated occurrences of the term "uka" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

matically

-0.89

mens

-0.83

shots

-0.74

shot

-0.72

hip

-0.72

izoph

-0.71

tons

-0.70

sc

-0.69

mental

-0.69

ror

-0.68

POSITIVE LOGITS

ichi

0.95

Cola

0.86

ovych

0.84

emia

0.82

enko

0.80

atu

0.76

uka

0.75

ernel

0.72

Å¡

0.71

Ã±a

0.71

Activations Density 0.019%

the name "Luka"

repeated occurrences of the term "uka" in various contexts

No Comments

No Known Activations