INDEX

Explanations

phrases related to exploration and investigation

oai_token-act-pair · gpt-3.5-turbo

terms related to measurement and quantities

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

 Gladiator

-0.82

 Brush

-0.81

 Blossom

-0.78

 Painter

-0.78

 Cousins

-0.77

 Rampage

-0.77

 Clouds

-0.77

 Zuckerberg

-0.76

 Exploration

-0.76

Shock

-0.76

POSITIVE LOGITS

ma

1.64

mi

1.54

mu

1.52

nic

1.49

li

1.46

cas

1.45

ja

1.45

sam

1.45

nu

1.45

il

1.45

Activations Density 0.108%

phrases related to exploration and investigation

terms related to measurement and quantities

No Comments

No Known Activations