INDEX

Explanations

patterns related to programming code, file paths or commands

oai_token-act-pair · gpt-3.5-turbo

alphanumeric symbols and construction-related terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inki

-0.76

clinton

-0.63

isSpecialOrderable

-0.62

ail

-0.62

theless

-0.62

awaru

-0.61

agher

-0.58

 thirds

-0.57

zik

-0.57

 spat

-0.55

POSITIVE LOGITS

 Sold

0.64

 Gleaming

0.63

flows

0.57

 leaps

0.56

Jiu

0.55

artifacts

0.54

 guiActiveUnfocused

0.53

 Engineers

0.53

 Expand

0.52

 Limits

0.52

Activations Density 1.297%

patterns related to programming code, file paths or commands

alphanumeric symbols and construction-related terminology

No Comments

No Known Activations