INDEX

Explanations

adjectives describing a remarkable or impressive quality

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

abet

-0.74

gard

-0.71

enser

-0.70

QB

-0.70

Va

-0.67

ħĭ

-0.67

gart

-0.66

aimon

-0.66

ettel

-0.65

hemy

-0.65

POSITIVE LOGITS

 striking

0.99

 strike

0.77

 resemb

0.74

 Jagu

0.73

 shifts

0.71

 struck

0.71

 Strikes

0.69

 strikes

0.68

 resemblance

0.68

 unions

0.67

Activations Density 0.005%

adjectives describing a remarkable or impressive quality

No Comments

No Known Activations