INDEX

Explanations

specific structures or patterns within text, such as sequences of numbers and characters

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 overhe

-0.73

Kin

-0.72

Blu

-0.71

netflix

-0.69

odynam

-0.68

 apparel

-0.68

geries

-0.66

 assum

-0.64

onse

-0.63

cdn

-0.62

POSITIVE LOGITS

 suffix

1.20

 aster

1.19

 denotes

1.17

 prefix

1.16

 denote

1.14

 notation

1.12

 comma

1.11

"+

1.10

"-

1.08

 initials

1.06

Activations Density 0.332%

specific structures or patterns within text, such as sequences of numbers and characters

No Comments

No Known Activations