INDEX

Explanations

occurrences of the text "SP" with varying activations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the abbreviation "SP" with varying contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

meal

-0.82

hran

-0.79

edin

-0.76

anchester

-0.73

tn

-0.68

oglu

-0.67

gebra

-0.63

tackle

-0.62

ornia

-0.61

Fr

-0.61

POSITIVE LOGITS

SP

3.70

SP

2.19

SPL

1.52

SPD

1.48

SPR

1.46

SM

1.40

RP

1.37

Sp

1.36

SL

1.35

SPI

1.34

Activations Density 0.014%

occurrences of the text "SP" with varying activations

occurrences of the abbreviation "SP" with varying contexts

No Comments

No Known Activations