INDEX

Explanations

locations and events described in news articles

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 positives

-0.71

 chore

-0.70

 probes

-0.68

 narciss

-0.67

 ticking

-0.67

 Reincarn

-0.66

 fingerprints

-0.66

 negotiator

-0.65

 smoot

-0.65

 sque

-0.64

POSITIVE LOGITS

 Springs

1.23

town

1.22

 Township

1.18

esville

1.18

 Valley

1.16

 Lakes

1.16

 Falls

1.16

 Creek

1.15

stown

1.14

minster

1.14

Activations Density 2.922%

locations and events described in news articles

No Comments

No Known Activations