INDEX

Explanations

special characters that are likely used for encoding or representing foreign characters

oai_token-act-pair · gpt-3.5-turbo

references to a specific name or term repeatedly mentioned in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

è¦ļéĨĴ

-0.87

ngth

-0.76

 imperson

-0.74

awaru

-0.70

jriwal

-0.70

ulas

-0.69

 tampering

-0.67

 takeoff

-0.67

 hatched

-0.67

iston

-0.66

POSITIVE LOGITS

CEPT

0.89

ãĥŃ

0.89

×ķ

0.88

ÙĨ

0.88

×Ļ×

0.86

Ĵ

0.85

×¨

0.84

VEN

0.84

BOOK

0.84

BLE

0.83

Activations Density 0.006%

special characters that are likely used for encoding or representing foreign characters

references to a specific name or term repeatedly mentioned in various contexts

No Comments

No Known Activations