INDEX

Explanations

phrases or clauses related to plans, intentions, or actions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating relationships or conditions involving "as"

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itiveness

-0.85

uca

-0.79

ingo

-0.67

ivas

-0.66

alog

-0.66

alogy

-0.65

opsis

-0.62

romy

-0.62

icus

-0.62

 Appearance

-0.59

POSITIVE LOGITS

 part

1.22

 punishment

1.11

 retribution

1.05

phy

1.01

 retaliation

0.97

 soon

0.96

 early

0.92

pired

0.91

 revenge

0.86

part

0.85

Activations Density 0.125%

phrases or clauses related to plans, intentions, or actions

phrases indicating relationships or conditions involving "as"

No Comments

No Known Activations