INDEX

Explanations

phrases indicating methods or approaches

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

usters

-0.84

è¦ļéĨĴ

-0.81

ĸļ

-0.70

 disappoint

-0.69

 fumes

-0.68

 subsequ

-0.67

 livest

-0.65

encia

-0.65

 misunder

-0.64

ropolitan

-0.64

POSITIVE LOGITS

fare

1.34

finding

1.20

point

1.18

ward

1.17

points

1.06

finder

1.03

forward

0.95

station

0.89

bill

0.87

WARD

0.82

Activations Density 0.043%

phrases indicating methods or approaches

No Comments

No Known Activations