INDEX

Explanations

instances of negativity or critical sentiment

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

 Nicarag

-0.74

 Arabian

-0.72

 EntityItem

-0.63

 Dimension

-0.62

 Millennium

-0.60

 Galile

-0.60

 PLoS

-0.60

 Heavenly

-0.60

 sodium

-0.59

 Territory

-0.58

POSITIVE LOGITS

outs

1.23

out

1.20

away

1.16

up

1.15

around

1.14

offs

1.13

through

1.13

back

1.11

ups

1.10

forward

1.07

Activations Density 0.030%

instances of negativity or critical sentiment

No Comments

No Known Activations