INDEX

Explanations

words related to rankings or listings

oai_token-act-pair · gpt-3.5-turbo

references to ranking lists

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

 DRAG

-0.67

 streng

-0.66

 rall

-0.64

wards

-0.63

 livestream

-0.62

 Towns

-0.60

arching

-0.59

icer

-0.59

hours

-0.58

 Sidd

-0.57

POSITIVE LOGITS

erv

0.86

geist

0.83

 Kissinger

0.82

erve

0.80

witz

0.79

erves

0.72

MSN

0.72

 Priority

0.71

priority

0.71

tery

0.68

Activations Density 0.042%

words related to rankings or listings

references to ranking lists

No Comments

No Known Activations