INDEX

Explanations

numbers combined with words related to events or actions

oai_token-act-pair · gpt-3.5-turbo

numeric values or statistics

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 administr

-0.90

 proble

-0.77

 newsp

-0.76

 challeng

-0.75

 manif

-0.73

 manifold

-0.72

 brid

-0.69

undai

-0.66

 federation

-0.66

 leve

-0.65

POSITIVE LOGITS

 ILCS

1.23

 SHARES

1.10

1.01

0.94

0.93

0.92

 Explicit

0.90

0.89

0.88

Activations Density 0.097%

numbers combined with words related to events or actions

numeric values or statistics

No Comments

No Known Activations