INDEX

Explanations

phrases related to lists or tracking items

oai_token-act-pair · gpt-3.5-turbo

structured lists and records of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ĨĴ

-0.76

 acknowled

-0.71

ruary

-0.68

 SERV

-0.66

issance

-0.64

 princ

-0.64

Posted

-0.63

ratulations

-0.63

 Marketable

-0.62

 lesson

-0.60

POSITIVE LOGITS

 frequencies

0.72

visor

0.70

Quality

0.69

Beh

0.68

crim

0.66

fts

0.65

 allerg

0.65

edin

0.64

rouse

0.64

orah

0.64

Activations Density 0.362%

phrases related to lists or tracking items

structured lists and records of information

No Comments

No Known Activations