INDEX

Explanations

instances of words beginning with 'Im' followed by a high activation score, potentially indicating a focus on imperialism

oai_token-act-pair · gpt-3.5-turbo

forms of the word "improve" or "implementation."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 notoriously

-0.74

 Fortune

-0.69

PAC

-0.69

 scarce

-0.68

 fortune

-0.66

 Bates

-0.66

 secondary

-0.65

RC

-0.65

 troubled

-0.65

 Coleman

-0.65

POSITIVE LOGITS

Im

3.62

im

1.45

Im

1.31

Em

1.30

IM

1.28

Imm

1.27

Imp

1.19

Ur

1.17

Inst

1.16

Inf

1.12

Activations Density 0.008%

instances of words beginning with 'Im' followed by a high activation score, potentially indicating a focus on imperialism

forms of the word "improve" or "implementation."

No Comments

No Known Activations