INDEX

Explanations

terms related to numerical offsets or adjustments

oai_token-act-pair · gpt-3.5-turbo

terms related to adjustments or compensations, especially in a numerical or financial 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Guilty

-0.75

 Prem

-0.66

Fel

-0.65

 Cult

-0.65

chel

-0.63

 Gran

-0.63

gent

-0.63

goers

-0.63

Commun

-0.62

 Commun

-0.61

POSITIVE LOGITS

 offset

3.75

 offsets

2.86

offset

2.27

Offset

1.91

 compensate

1.40

 compensated

1.30

 undercut

1.19

 outwe

1.16

 outweigh

1.12

 counteract

1.11

Activations Density 0.013%

terms related to numerical offsets or adjustments

terms related to adjustments or compensations, especially in a numerical or financial context

No Comments

No Known Activations