INDEX

Explanations

The neuron is looking for terms related to proprietary information, such as "proprietary challenge-response authentication scheme."

oai_token-act-pair · gpt-3.5-turbo

terms related to proprietary and copyrighted content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

OPA

-0.84

stad

-0.82

aper

-0.75

othy

-0.75

orah

-0.73

SEE

-0.73

alos

-0.72

ohan

-0.71

den

-0.69

older

-0.69

POSITIVE LOGITS

 proprietary

0.95

 interests

0.82

 formulas

0.80

 materials

0.80

 protocols

0.78

 blends

0.77

 secrets

0.77

ĨĴ

0.76

 jargon

0.76

 terminology

0.76

Activations Density 0.018%

The neuron is looking for terms related to proprietary information, such as "proprietary challenge-response authentication scheme."

terms related to proprietary and copyrighted content

No Comments

No Known Activations