INDEX

Explanations

phrases related to possession, description, or attribution

oai_token-act-pair · gpt-3.5-turbo

legal terminology or references in a specific context

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

 relaxing

-0.62

 stranded

-0.62

 corpus

-0.61

boo

-0.60

 Millennium

-0.59

 trapped

-0.57

rip

-0.55

illy

-0.55

 pensions

-0.54

 squeeze

-0.54

POSITIVE LOGITS

 thereof

1.98

oway

1.63

language

1.55

 circumstance

1.52

techn

1.23

CN

1.13

purpose

1.10

istrate

1.07

 thereto

1.07

major

1.02

Activations Density 0.034%

phrases related to possession, description, or attribution

legal terminology or references in a specific context

No Comments

No Known Activations