INDEX

Explanations

actions that users can take within applications, such as checking, creating, sharing, or streaming content

oai_token-act-pair · gpt-3.5-turbo

phrases indicating user capabilities or actions

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

rant

-0.72

Moz

-0.72

 Irving

-0.69

 sterling

-0.64

 Principles

-0.64

fell

-0.64

dad

-0.62

 Cous

-0.62

 Confeder

-0.62

 Derrick

-0.61

POSITIVE LOGITS

 customize

1.14

 optionally

1.11

 manually

0.91

use

0.90

 manipulate

0.88

 utilize

0.86

 specify

0.83

 browse

0.83

 interact

0.83

 modify

0.82

Activations Density 0.262%

actions that users can take within applications, such as checking, creating, sharing, or streaming content

phrases indicating user capabilities or actions

No Comments

No Known Activations