INDEX

Explanations

instances of the word "Sa" followed by a numerical value, potentially a name or abbreviation

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the name "Sa"

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

lessly

-0.78

papers

-0.77

ships

-0.72

tics

-0.71

mercial

-0.66

å§«

-0.64

breaks

-0.64

 Establishment

-0.64

INGTON

-0.63

liness

-0.62

POSITIVE LOGITS

igon

1.08

pling

1.04

vers

1.02

plings

1.01

plin

0.98

izo

0.97

Ga

0.97

uten

0.95

qqa

0.94

iva

0.92

Activations Density 0.018%

instances of the word "Sa" followed by a numerical value, potentially a name or abbreviation

repeated mentions of the name "Sa"

No Comments

No Known Activations