INDEX

Explanations

references to performances or events

oai_token-act-pair · gpt-3.5-turbo

the repeated mention of the term "Per" related to performances or events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

INESS

-0.67

 pneum

-0.65

ä

-0.61

 Lazarus

-0.60

 braces

-0.60

 Raleigh

-0.59

 Tycoon

-0.58

RIS

-0.58

 alone

-0.58

 dwar

-0.57

POSITIVE LOGITS

malink

1.41

uvian

1.37

ipher

1.30

forming

1.30

pet

1.28

cussion

1.27

mission

1.26

missions

1.25

secution

1.25

imeter

1.24

Activations Density 0.020%

references to performances or events

the repeated mention of the term "Per" related to performances or events

No Comments

No Known Activations