INDEX

Explanations

locations or specific areas of interest, potentially related to a secretive or restricted place

oai_token-act-pair · gpt-3.5-turbo

terms related to geographic and structural classifications, often in a bureaucratic or organizational context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

iage

-0.75

enegger

-0.74

issance

-0.73

sterdam

-0.71

paces

-0.70

nesday

-0.70

ashington

-0.68

iants

-0.68

uckland

-0.67

lisher

-0.67

POSITIVE LOGITS

 Specific

0.79

 Zero

0.75

Maker

0.71

 Selection

0.69

 Pieces

0.69

 Breaker

0.69

 Objects

0.68

 Eight

0.68

 Size

0.68

 Unknown

0.68

Activations Density 0.241%

locations or specific areas of interest, potentially related to a secretive or restricted place

terms related to geographic and structural classifications, often in a bureaucratic or organizational context

No Comments

No Known Activations