INDEX

Explanations

anti-<specific group or issue> sentiments

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

ĸļ

-0.74

 Sins

-0.68

 Pelicans

-0.64

 resid

-0.64

ãĤ¼ãĤ¦ãĤ¹

-0.63

 Gorge

-0.63

 Legends

-0.63

 Halls

-0.62

 Tycoon

-0.61

 Taste

-0.61

POSITIVE LOGITS

abortion

1.21

establishment

1.16

war

1.10

hero

0.97

strate

0.93

overty

0.90

immigrant

0.89

Semitic

0.88

jud

0.87

inflammatory

0.87

Activations Density 0.316%

anti-<specific group or issue> sentiments

No Comments

No Known Activations