INDEX

Explanations

references to locations or positions, particularly those near the end or bottom

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "near" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 CODE

-0.69

OUP

-0.66

Zin

-0.63

 behav

-0.63

PLA

-0.62

 lawy

-0.60

 assurances

-0.59

 Doodle

-0.59

Zy

-0.59

 Mahjong

-0.59

POSITIVE LOGITS

ricular

0.80

ctic

0.77

clock

0.75

 periphery

0.74

astern

0.73

 horizon

0.70

end

0.70

interstitial

0.70

 crowded

0.70

ĸļ

0.68

Activations Density 0.091%

references to locations or positions, particularly those near the end or bottom

occurrences of the word "near" in various contexts

No Comments

No Known Activations