INDEX

Explanations

statements expressing feelings or opinions, often with words like "happy", "disappointing", and "pleased"

oai_token-act-pair · gpt-3.5-turbo

expressions of emotions and opinions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãĥ¯ãĥ³

-0.77

apeshifter

-0.74

definition

-0.73

spec

-0.71

æŃ¦

-0.66

ixt

-0.66

esm

-0.66

 arsen

-0.65

defined

-0.64

illac

-0.64

POSITIVE LOGITS

 glad

1.44

 thankful

1.39

 thanking

1.31

 applaud

1.29

 congratulate

1.27

 pity

1.27

 hope

1.25

 rejoice

1.24

 grateful

1.24

 hopeful

1.22

Activations Density 0.720%

statements expressing feelings or opinions, often with words like "happy", "disappointing", and "pleased"

expressions of emotions and opinions

No Comments

No Known Activations