INDEX

Explanations

the word "signal" with varying activations

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.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

sect

-0.75

spell

-0.70

erenn

-0.66

yright

-0.66

ttes

-0.65

venge

-0.65

shop

-0.64

sm

-0.63

uum

-0.62

ositories

-0.61

POSITIVE LOGITS

 emanating

0.96

 amplification

0.90

 signals

0.86

 handlers

0.86

 signal

0.85

 emitted

0.84

 strength

0.83

 emitting

0.83

 handler

0.83

 propagation

0.83

Activations Density 0.045%

the word "signal" with varying activations

No Comments

No Known Activations