INDEX

Explanations

phrases indicating contrast or exception within a sentence

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate negation or limitations in assertions

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

ilage

-0.75

iol

-0.72

 Circuit

-0.63

iology

-0.61

iov

-0.61

 reperto

-0.60

 Royale

-0.60

ulas

-0.59

 manif

-0.59

 Union

-0.59

POSITIVE LOGITS

 necessarily

0.88

 whatsoever

0.79

nor

0.75

 bothered

0.71

necess

0.69

 condone

0.68

 endorse

0.67

xus

0.65

ndra

0.65

icably

0.64

Activations Density 0.044%

phrases indicating contrast or exception within a sentence

phrases that indicate negation or limitations in assertions

No Comments

No Known Activations