INDEX

Explanations

the word "possible" followed by a high activation number

oai_token-act-pair · gpt-3.5-turbo

phrases that express possibilities or potential outcomes

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

bane

-0.79

cipl

-0.77

bey

-0.77

phas

-0.75

waters

-0.70

zac

-0.70

ophone

-0.70

eye

-0.67

roma

-0.67

cig

-0.66

POSITIVE LOGITS

Â¶

0.77

FORE

0.73

NK

0.70

 exit

0.70

terday

0.69

ossibility

0.65

 partName

0.65

 understatement

0.64

lly

0.62

 nevertheless

0.61

Activations Density 0.021%

the word "possible" followed by a high activation number

phrases that express possibilities or potential outcomes

No Comments

No Known Activations