INDEX

Explanations

adjective-noun pairs and specific terms related to technology/security

oai_token-act-pair · gpt-3.5-turbo

terms related to measurement and evaluation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Accountability

-0.77

 Salvation

-0.76

 Grimes

-0.74

 Millennium

-0.72

 Spur

-0.71

 Hammond

-0.69

 Darling

-0.67

 Enterprise

-0.65

 Awakening

-0.65

 Abyss

-0.64

POSITIVE LOGITS

lest

0.96

ativity

0.94

Ð¾

0.94

itial

0.94

illary

0.93

anc

0.92

umin

0.91

ittance

0.91

onal

0.90

ero

0.89

Activations Density 0.404%

adjective-noun pairs and specific terms related to technology/security

terms related to measurement and evaluation

No Comments

No Known Activations