INDEX

Explanations

numeric patterns and step-related instructions

oai_token-act-pair · gpt-3.5-turbo

numerical values and their corresponding identifiers or classifications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.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

chy

-0.81

 minded

-0.77

pers

-0.69

etimes

-0.68

itness

-0.67

minded

-0.66

istical

-0.62

Ward

-0.62

ement

-0.61

ivities

-0.61

POSITIVE LOGITS

 onwards

0.82

 onward

0.72

 Parables

0.70

 Presents

0.68

 Ships

0.62

Aux

0.62

ĨĴ

0.62

 Forth

0.62

 Puzzles

0.61

 =================

0.61

Activations Density 0.065%

numeric patterns and step-related instructions

numerical values and their corresponding identifiers or classifications

No Comments

No Known Activations