INDEX

Explanations

references to specific locations or events, particularly in journalistic content

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

 majesty

-0.67

 expire

-0.67

 discrim

-0.67

 hops

-0.65

 registry

-0.64

 plants

-0.62

 fung

-0.62

 retard

-0.61

 orche

-0.61

 untreated

-0.61

POSITIVE LOGITS

Reuters

1.27

Thom

1.22

CBS

1.07

AFP

1.00

CNN

0.97

credit

0.96

CN

0.94

Associated

0.92

Bloomberg

0.92

AP

0.90

Activations Density 0.049%

references to specific locations or events, particularly in journalistic content

No Comments

No Known Activations