INDEX

Explanations

phrases related to providing instructions or suggestions

oai_token-act-pair · gpt-3.5-turbo

references to guidance or instructions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ells

-0.79

otrop

-0.70

ols

-0.70

 Roses

-0.69

Ve

-0.68

Sus

-0.67

itol

-0.66

ãĥ£

-0.65

hma

-0.64

POSE

-0.64

POSITIVE LOGITS

 guidance

1.18

 counselors

1.03

 counselor

1.01

 Guid

0.85

 guideline

0.84

 guiding

0.83

 advising

0.83

 guidelines

0.82

idelines

0.81

 guides

0.79

Activations Density 0.010%

phrases related to providing instructions or suggestions

references to guidance or instructions

No Comments

No Known Activations