INDEX

Explanations

references to political events and government actions

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

Quantity

-0.80

 ..."

-0.71

trop

-0.69

 crappy

-0.68

 decency

-0.68

 ______

-0.66

 shitty

-0.65

Fuck

-0.65

nig

-0.65

 crap

-0.64

POSITIVE LOGITS

 meanwhile

1.06

 reportedly

1.05

 prompted

1.03

 also

1.02

 however

0.92

 bolstered

0.87

 additionally

0.85

 spokeswoman

0.85

 furthermore

0.84

 spokesman

0.84

Activations Density 9.481%

references to political events and government actions

No Comments

No Known Activations