INDEX

Explanations

references to specific entities or terms like "Dash" in the context of a news story

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

urally

-0.75

zzo

-0.73

ãĥĩãĤ£

-0.71

ervation

-0.64

glers

-0.63

ppo

-0.63

>>>>>>>>

-0.62

cffff

-0.61

cientious

-0.59

chard

-0.59

POSITIVE LOGITS

boards

1.13

board

1.03

cam

1.00

dash

0.93

lane

0.91

Dash

0.80

ription

0.75

 dash

0.73

hawk

0.73

robe

0.72

Activations Density 0.011%

references to specific entities or terms like "Dash" in the context of a news story

No Comments

No Known Activations