INDEX

Explanations

information or references to op-eds

oai_token-act-pair · gpt-3.5-turbo

instances of the term "op-ed" and related writing formats

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

¶ħ

-0.77

 Galileo

-0.77

 Arabia

-0.73

auga

-0.71

ãĤ£

-0.70

Suc

-0.67

ãĥ¯ãĥ³

-0.67

KDE

-0.66

wagen

-0.66

 Macintosh

-0.65

POSITIVE LOGITS

iates

1.06

onent

1.04

osite

1.00

orter

0.98

iate

0.97

inion

0.91

rog

0.90

ressor

0.89

osition

0.88

elt

0.88

Activations Density 0.009%

information or references to op-eds

instances of the term "op-ed" and related writing formats

No Comments

No Known Activations