INDEX

Explanations

negative phrases or expressions that indicate failure or decline

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

 Cyrus

-0.77

 fireworks

-0.77

Mub

-0.72

 Pengu

-0.72

ado

-0.72

ysis

-0.71

FML

-0.71

 Thrones

-0.69

 Tsukuyomi

-0.69

 Clash

-0.69

POSITIVE LOGITS

based

1.29

forward

1.26

burning

1.23

backed

1.14

roots

1.12

centered

1.10

centric

1.07

covered

1.07

related

1.05

bound

1.05

Activations Density 0.037%

negative phrases or expressions that indicate failure or decline

No Comments

No Known Activations