INDEX

Explanations

the presence of numerical spans or ranges

oai_token-act-pair · gpt-3.5-turbo

instances of the word "span" and its variations, indicating a focus on duration or extent

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

illard

-0.75

tical

-0.68

eers

-0.68

ysis

-0.67

tics

-0.64

 aspir

-0.64

 assistants

-0.64

 heirs

-0.63

ertodd

-0.63

sie

-0.61

POSITIVE LOGITS

ning

0.93

ming

0.91

 span

0.91

span

0.90

 spanning

0.87

eworks

0.84

itime

0.84

ner

0.79

agy

0.79

ners

0.78

Activations Density 0.013%

the presence of numerical spans or ranges

instances of the word "span" and its variations, indicating a focus on duration or extent

No Comments

No Known Activations