INDEX

Explanations

phrases related to specific titles or names

oai_token-act-pair · gpt-3.5-turbo

references to specific titles or franchises

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mble

-0.87

holes

-0.75

kaya

-0.73

nai

-0.69

oulos

-0.68

keye

-0.68

ascript

-0.67

hement

-0.67

 disg

-0.66

OPLE

-0.66

POSITIVE LOGITS

 Month

0.89

 Updated

0.83

ime

0.83

 Year

0.82

 Gleaming

0.78

 month

0.74

val

0.73

 Testament

0.71

 week

0.71

 Night

0.70

Activations Density 0.019%

phrases related to specific titles or names

references to specific titles or franchises

No Comments

No Known Activations