INDEX

Explanations

the abbreviation "NA" with different numerical values

oai_token-act-pair · gpt-3.5-turbo

references to a specific entity or concept denoted by "NA"

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

ienced

-0.75

======

-0.72

papers

-0.71

tons

-0.71

lift

-0.69

birds

-0.68

loop

-0.67

nuts

-0.67

hold

-0.67

starter

-0.66

POSITIVE LOGITS

ZI

1.30

ACP

1.11

FU

0.94

WD

0.93

VE

0.91

NA

0.90

BLE

0.89

BER

0.88

VAL

0.84

ISS

0.83

Activations Density 0.012%

the abbreviation "NA" with different numerical values

references to a specific entity or concept denoted by "NA"

No Comments

No Known Activations