INDEX

Explanations

occurrences of the word "new"

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

 ..........

-0.71

 Brach

-0.70

dod

-0.60

 mishand

-0.59

 peanuts

-0.59

ILA

-0.59

 sidx

-0.59

ONSORED

-0.58

 Cele

-0.57

BILITY

-0.57

POSITIVE LOGITS

bies

1.38

bie

1.29

foundland

1.24

found

1.11

 Zealand

1.06

tons

1.05

 arrivals

0.98

castle

0.94

sc

0.92

York

0.92

Activations Density 0.082%

occurrences of the word "new"

No Comments

No Known Activations