INDEX

Explanations

negative sentiments or descriptors, particularly words with 'unf' prefixed

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of being "unfunny" or lacking humor

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 gorilla

-0.69

 Disorders

-0.65

 initials

-0.65

 Bots

-0.63

 Jordan

-0.63

 Floor

-0.62

Kry

-0.62

 Madness

-0.61

 speakers

-0.61

 Genie

-0.61

POSITIVE LOGITS

athom

1.69

riend

1.53

avour

1.43

ortun

1.43

ortunately

1.41

ashion

1.29

aith

1.25

rozen

1.25

itting

1.22

requ

1.21

Activations Density 0.010%

negative sentiments or descriptors, particularly words with 'unf' prefixed

terms related to the concept of being "unfunny" or lacking humor

No Comments

No Known Activations