INDEX

Explanations

references to a specific entity or term, likely the word "Com"

oai_token-act-pair · gpt-3.5-turbo

references to "Com" in various contexts, likely indicating a focus on organizations or commissions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Kubrick

-0.86

 harbor

-0.74

 Icelandic

-0.73

 Mara

-0.69

 halfway

-0.67

 dwar

-0.65

 Keynes

-0.64

 Kund

-0.63

 Michaels

-0.62

BDS

-0.61

POSITIVE LOGITS

ptroller

1.26

fortable

1.18

pletion

1.15

pleting

1.13

mented

1.11

ittee

1.08

plex

1.05

intern

1.03

rade

1.01

anooga

1.01

Activations Density 0.013%

references to a specific entity or term, likely the word "Com"

references to "Com" in various contexts, likely indicating a focus on organizations or commissions

No Comments

No Known Activations