INDEX

Explanations

terms related to discrimination and restrictions on individuals based on their identity or past offenses

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

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.03

1:0.04

2:0.09

3:0.17

4:0.03

5:0.03

6:0.10

7:0.10

8:0.05

9:0.08

10:0.09

11:0.16

Negative Logits

osc

-0.96

 Puzz

-0.94

 disapp

-0.93

 dwind

-0.92

��極

-0.91

 horm

-0.91

Tube

-0.89

 embr

-0.89

wagon

-0.89

wagen

-0.89

POSITIVE LOGITS

aram

0.98

ateral

0.95

 Fram

0.90

Nik

0.89

ITNESS

0.88

allas

0.88

ivable

0.86

ociate

0.86

VICE

0.86

 martial

0.86

Activations Density 0.002%

terms related to discrimination and restrictions on individuals based on their identity or past offenses

No Comments

No Known Activations