INDEX

Explanations

specific entities or names, potentially related to events or locations

oai_token-act-pair · gpt-3.5-turbo

references to specific organizations or named entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

luaj

-0.87

ãĥ¯ãĥ³

-0.86

hole

-0.83

rote

-0.81

ahime

-0.81

intosh

-0.78

estern

-0.78

RAFT

-0.78

holes

-0.77

vernment

-0.76

POSITIVE LOGITS

 Goods

0.76

 SEAL

0.65

OPLE

0.64

 poppy

0.64

born

0.63

 Sons

0.62

 Malfoy

0.62

 Nieto

0.62

 Debt

0.62

beam

0.61

Activations Density 0.030%

specific entities or names, potentially related to events or locations

references to specific organizations or named entities

No Comments

No Known Activations