INDEX

Explanations

specific technical terms or labels related to different iterations or versions of a certain product

oai_token-act-pair · gpt-3.5-turbo

specific model and product names in a technical or entertainment context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

theless

-0.85

 Collider

-0.77

NESS

-0.75

film

-0.70

Cand

-0.66

Madison

-0.65

grounds

-0.65

 Garry

-0.65

ball

-0.64

DEN

-0.64

POSITIVE LOGITS

ensible

1.05

olon

0.88

anting

0.88

olphins

0.83

ilon

0.83

retched

0.82

urat

0.82

Mk

0.82

oppable

0.80

osterone

0.80

Activations Density 0.025%

specific technical terms or labels related to different iterations or versions of a certain product

specific model and product names in a technical or entertainment context

No Comments

No Known Activations