INDEX

Explanations

phrases related to permissions or reprints

oai_token-act-pair · gpt-3.5-turbo

terms related to publishing and reprinting content

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

Â¯Â¯Â¯Â¯

-0.77

hart

-0.73

rera

-0.66

holes

-0.64

apo

-0.64

shore

-0.62

gamer

-0.61

omen

-0.61

erb

-0.60

Â¯Â¯

-0.60

POSITIVE LOGITS

 reprint

1.06

 copies

0.92

lishing

0.83

rint

0.82

ensing

0.81

ãĤ´

0.81

 excerpts

0.79

DragonMagazine

0.79

letters

0.79

 reuse

0.79

Activations Density 0.067%

phrases related to permissions or reprints

terms related to publishing and reprinting content

No Comments

No Known Activations