INDEX

Explanations

terms related to a specific word "rat" with differing numerical values

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "rat"

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

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

 Matter

-0.82

Cue

-0.66

ription

-0.64

 marrow

-0.60

Pry

-0.60

ozy

-0.60

 Blaz

-0.60

 Cold

-0.60

 Wizards

-0.60

 Month

-0.58

POSITIVE LOGITS

itude

0.97

acus

0.95

oday

0.95

tered

0.93

itudes

0.91

igraph

0.90

hetic

0.89

tenance

0.87

itism

0.87

istical

0.87

Activations Density 0.020%

terms related to a specific word "rat" with differing numerical values

mentions of the term "rat"

No Comments

No Known Activations