INDEX

Explanations

words related to automatic processes

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate automated processes or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Straw

-0.80

 Emin

-0.76

ĸļ

-0.76

 Mouth

-0.76

rug

-0.72

ergus

-0.72

ador

-0.71

wife

-0.70

gerald

-0.69

 Prescott

-0.69

POSITIVE LOGITS

 populate

0.94

 induct

0.90

 detects

0.86

 detect

0.86

 migrate

0.84

 aspir

0.82

untarily

0.81

 indemn

0.81

 enrolled

0.81

 generated

0.81

Activations Density 0.011%

words related to automatic processes

phrases that indicate automated processes or actions

No Comments

No Known Activations