INDEX

Explanations

the word "zero" preceded or followed by dashes or numbers, indicating a mathematical or policy context

oai_token-act-pair · gpt-3.5-turbo

concepts related to zero tolerance policies

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

rough

-0.80

Downloadha

-0.71

saw

-0.70

nikov

-0.70

RAW

-0.65

ammad

-0.64

aws

-0.64

WithNo

-0.64

regor

-0.63

oured

-0.63

POSITIVE LOGITS

 tolerance

1.09

sum

0.93

 percent

0.88

MQ

0.87

pointer

0.80

 volts

0.79

 gravity

0.78

onday

0.76

ed

0.74

 Hedge

0.73

Activations Density 0.038%

the word "zero" preceded or followed by dashes or numbers, indicating a mathematical or policy context

concepts related to zero tolerance policies

No Comments

No Known Activations