INDEX

Explanations

phrases related to a specific subject or topic being questioned or discussed

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

 landmark

-0.63

 anniversary

-0.61

col

-0.58

}}

-0.57

coming

-0.56

 Previously

-0.56

listed

-0.56

 particularly

-0.55

 suprem

-0.55

bra

-0.55

POSITIVE LOGITS

 merely

1.26

 simply

1.05

 concentrate

1.02

 purely

0.87

Instead

0.80

 relying

0.80

foc

0.78

 instead

0.75

 solely

0.74

 focus

0.74

Activations Density 2.573%

phrases related to a specific subject or topic being questioned or discussed

No Comments

No Known Activations