INDEX

Explanations

phrases that highlight the interests of various groups or individuals

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_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.02

2:0.07

3:0.06

4:0.13

5:0.03

6:0.03

7:0.29

8:0.04

9:0.04

10:0.12

11:0.10

Negative Logits

 recalls

-1.38

 tracks

-1.37

pez

-1.34

ゴン

-1.33

ーン

-1.32

 makeshift

-1.32

was

-1.30

steps

-1.29

 spiked

-1.28

earable

-1.27

POSITIVE LOGITS

ibur

1.77

 shareholders

1.75

 shareholder

1.59

 preservation

1.49

orsche

1.48

 maximizing

1.45

 commerce

1.42

 warr

1.42

 preserving

1.40

 protecting

1.39

Activations Density 0.008%

phrases that highlight the interests of various groups or individuals

No Comments

No Known Activations