INDEX

Explanations

the word "don't."

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

0:0.06

1:0.07

2:0.07

3:0.08

4:0.08

5:0.08

6:0.10

7:0.08

8:0.07

9:0.08

10:0.09

11:0.08

Negative Logits

>:

-1.99

etheless

-1.93

ylum

-1.88

cknow

-1.76

idth

-1.74

lehem

-1.70

nesday

-1.69

 livest

-1.62

adelphia

-1.60

bidden

-1.60

POSITIVE LOGITS

 pharmacies

1.79

Cs

1.76

 Stores

1.59

trace

1.57

Footnote

1.57

Store

1.51

Contents

1.50

 Forms

1.49

fect

1.47

ingu

1.46

Activations Density 0.000%

the word "don't."

No Comments

No Known Activations