INDEX

Explanations

words related to unusual or strange concepts or events

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

ptive

-0.87

ptives

-0.84

FORE

-0.78

vation

-0.77

ignty

-0.74

cussion

-0.70

aders

-0.70

HCR

-0.69

ailable

-0.67

ILA

-0.67

POSITIVE LOGITS

ness

1.36

nesses

1.19

ly

1.15

est

1.01

ed

0.97

oes

0.97

ety

0.95

os

0.92

er

0.89

ening

0.88

Activations Density 0.034%

words related to unusual or strange concepts or events

No Comments

No Known Activations