INDEX

Explanations

statements about physical abuse

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

majority

-0.61

°

-0.60

ories

-0.59

ģĸ

-0.58

 Coalition

-0.58

 Siege

-0.56

heny

-0.55

profits

-0.55

 Centers

-0.54

ĨĴ

-0.53

POSITIVE LOGITS

selves

1.10

atic

1.09

atically

1.08

self

1.01

 personally

0.97

lees

0.91

adows

0.91

 selves

0.85

andering

0.83

soever

0.82

Activations Density 3.208%

statements about physical abuse

No Comments

No Known Activations