INDEX

Explanations

addresses or locations based on street names and numbers

oai_token-act-pair · gpt-3.5-turbo

ordinal indicators in numerical contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

esc

-0.71

Statistics

-0.65

 Canaver

-0.61

instein

-0.60

orter

-0.60

etting

-0.57

housing

-0.56

 bleach

-0.56

rawdownloadcloneembedreportprint

-0.56

ombo

-0.55

POSITIVE LOGITS

ousand

1.10

 Amendment

1.01

 Avenue

0.97

 Street

0.95

 percentile

0.88

 Degree

0.87

 Floor

0.87

 Century

0.85

 Percent

0.84

 Anniversary

0.84

Activations Density 0.029%

addresses or locations based on street names and numbers

ordinal indicators in numerical contexts

No Comments

No Known Activations