INDEX

Explanations

patterns related to suffixes or endings of words

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

torch.float32

Hook Point

blocks.8.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

 DRAG

-0.69

ailability

-0.66

 substance

-0.64

ãĥīãĥ©ãĤ´ãĥ³

-0.62

llah

-0.62

PLIC

-0.61

iannopoulos

-0.61

 colleges

-0.61

 scant

-0.59

 concess

-0.59

POSITIVE LOGITS

kamp

0.92

furt

0.91

adder

0.89

wagen

0.86

 sidx

0.77

tsy

0.77

erella

0.73

fried

0.73

velt

0.69

bourg

0.68

Activations Density 0.018%

patterns related to suffixes or endings of words

No Comments

No Known Activations