INDEX

Explanations

the word "place".

oai_token-act-pair · gemini-1.5-flash Triggered by @johnny

locations or settings mentioned in the text

oai_token-act-pair · gpt-3.5-turbo

references to locations or places

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

rd

-0.70

olyn

-0.69

icer

-0.68

TAG

-0.65

ancers

-0.64

quel

-0.63

irst

-0.62

jin

-0.61

onder

-0.61

ÄŁ

-0.60

POSITIVE LOGITS

bos

1.14

holders

0.95

holder

0.83

 where

0.75

upon

0.74

 holder

0.74

places

0.71

ername

0.70

ħ

0.70

abouts

0.68

Activations Density 0.043%

the word "place".

locations or settings mentioned in the text

references to locations or places

No Comments

No Known Activations