INDEX

Explanations

phrases related to performing a search or locating something

oai_token-act-pair · gpt-3.5-turbo

references to searches or queries relating to locating information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

verning

-0.79

 Prot

-0.72

 Gavin

-0.71

ufact

-0.71

KI

-0.67

legates

-0.66

 horm

-0.66

volt

-0.66

 ceremonial

-0.65

¬¼

-0.64

POSITIVE LOGITS

 searches

2.01

 search

1.98

search

1.91

 Search

1.82

Search

1.81

 searching

1.79

 searched

1.75

 sear

1.55

 Sear

1.48

earchers

1.43

Activations Density 0.574%

phrases related to performing a search or locating something

references to searches or queries relating to locating information

No Comments

No Known Activations