INDEX

Explanations

instances where someone emphasizes a particular point or action they want to convey

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

thia

-0.62

assis

-0.62

Fram

-0.60

CLOSE

-0.59

heter

-0.58

Mania

-0.58

pac

-0.57

phrine

-0.57

bour

-0.57

rina

-0.57

POSITIVE LOGITS

 sure

1.23

hift

1.10

 sense

0.89

 landfall

0.86

 adjustments

0.83

 excuses

0.83

 strides

0.83

 headlines

0.81

URE

0.80

 decisions

0.78

Activations Density 0.799%

instances where someone emphasizes a particular point or action they want to convey

No Comments

No Known Activations