INDEX

Explanations

names of individuals

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

kefeller

-0.79

undo

-0.79

ype

-0.76

enhagen

-0.72

ypes

-0.72

raltar

-0.71

 sidx

-0.71

ozy

-0.69

vernment

-0.69

emetery

-0.68

POSITIVE LOGITS

 herself

1.71

pher

1.15

 Louise

1.12

 Anne

1.10

 Marie

1.09

 husband

1.07

 maid

1.04

 hijab

1.03

 vagina

1.03

athed

1.03

Activations Density 3.575%

names of individuals

No Comments

No Known Activations