INDEX

Explanations

references to specific dates or time and entities or objects

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Dixon

-0.74

 Slaughter

-0.63

 Sussex

-0.62

Spr

-0.61

 Showdown

-0.61

prising

-0.60

Fry

-0.60

mson

-0.59

nell

-0.59

 Spectre

-0.59

POSITIVE LOGITS

is

1.01

is

0.93

IS

0.78

 governs

0.77

 exists

0.75

isd

0.75

 denotes

0.74

Is

0.74

 guid

0.73

 arises

0.73

Activations Density 0.107%

references to specific dates or time and entities or objects

No Comments

No Known Activations