INDEX

Explanations

texts indicating a continuation or more information coming up

oai_token-act-pair · gpt-3.5-turbo

instances of ellipsis and other punctuation markings used for emphasis or continuation

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

ally

-0.71

ratulations

-0.70

hetto

-0.63

istically

-0.59

 attest

-0.58

 stresses

-0.58

adier

-0.57

gress

-0.56

äºº

-0.56

lor

-0.56

POSITIVE LOGITS

 âĢİ

0.87

BUT

0.77

 Appears

0.76

ļéĨĴ

0.75

wait

0.71

 Continue

0.69

ahime

0.69

 Written

0.67

until

0.67

Draft

0.66

Activations Density 0.069%

texts indicating a continuation or more information coming up

instances of ellipsis and other punctuation markings used for emphasis or continuation

No Comments

No Known Activations