INDEX

Explanations

terms related to the concept of 'concepts'

oai_token-act-pair · gpt-3.5-turbo

variations of the root word "cept," indicating concepts related to understanding, perception, or acceptance

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

 reconc

-0.78

 slaughtered

-0.74

 intest

-0.68

 swear

-0.65

Ń·

-0.65

©¶æ

-0.64

 tenancy

-0.63

 dining

-0.62

 reconciliation

-0.62

 marrow

-0.61

POSITIVE LOGITS

ional

1.41

ible

1.34

icons

1.28

ibility

1.27

ibles

1.26

ical

1.24

ual

1.24

acles

1.23

ually

1.23

ively

1.22

Activations Density 0.020%

terms related to the concept of 'concepts'

variations of the root word "cept," indicating concepts related to understanding, perception, or acceptance

No Comments

No Known Activations