INDEX

Explanations

phrases related to abstract concepts or symbolic representations

oai_token-act-pair · gpt-3.5-turbo

references to "lines" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

CVE

-0.79

CLASSIFIED

-0.78

ãĥ¼ãĥĨ

-0.77

rontal

-0.75

Rot

-0.74

KNOWN

-0.74

Assembly

-0.71

HL

-0.68

ADS

-0.66

ILD

-0.65

POSITIVE LOGITS

pace

1.01

mith

1.00

ystem

0.93

lines

0.92

omething

0.89

 lines

0.88

heet

0.86

poon

0.86

erv

0.82

hops

0.82

Activations Density 0.015%

phrases related to abstract concepts or symbolic representations

references to "lines" in various contexts

No Comments

No Known Activations