INDEX

Explanations

instances where someone disagrees, criticizes, or takes offense with something

oai_token-act-pair · gpt-3.5-turbo

phrases related to expressing objection or disagreement

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

function

-0.70

Phi

-0.68

itialized

-0.65

onder

-0.65

ctrl

-0.64

 Transform

-0.64

 Greenwich

-0.63

 Fall

-0.62

strength

-0.61

ayne

-0.61

POSITIVE LOGITS

 sarcast

0.85

 criticizing

0.80

 condem

0.78

 misrepresent

0.78

 dispar

0.78

UGH

0.77

 debunk

0.77

 approving

0.77

 spoof

0.76

 criticisms

0.76

Activations Density 0.193%

instances where someone disagrees, criticizes, or takes offense with something

phrases related to expressing objection or disagreement

No Comments

No Known Activations