INDEX

Explanations

words related to written communication, specifically letters addressed to someone

oai_token-act-pair · gpt-3.5-turbo

instances of the word "to" in various contexts related to letters or written correspondence

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

 deductions

-0.76

nels

-0.73

 bets

-0.71

ahime

-0.67

 wastes

-0.67

apter

-0.63

tions

-0.63

iHUD

-0.62

 cycles

-0.61

unning

-0.60

POSITIVE LOGITS

 accompany

1.06

 commemorate

0.98

wered

0.95

 clarify

0.94

pless

0.81

 ensure

0.81

 indicate

0.80

 signify

0.78

 address

0.78

 promote

0.75

Activations Density 0.169%

words related to written communication, specifically letters addressed to someone

instances of the word "to" in various contexts related to letters or written correspondence

No Comments

No Known Activations