INDEX

Explanations

phrases related to organization or sequence of events

oai_token-act-pair · gpt-3.5-turbo

references to sequential or upcoming events and topics

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

ceive

-0.66

 theirs

-0.65

anism

-0.63

 doorstep

-0.61

ozo

-0.61

internal

-0.60

every

-0.60

dom

-0.60

 Himself

-0.59

 pestic

-0.59

POSITIVE LOGITS

quickShipAvailable

0.89

 flaw

0.79

 drawback

0.79

 caveat

0.78

 question

0.77

 irony

0.74

 thing

0.73

question

0.71

 downside

0.71

 notable

0.70

Activations Density 0.149%

phrases related to organization or sequence of events

references to sequential or upcoming events and topics

No Comments

No Known Activations