INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L7_Hcat_z_lr1.20e-03_l11.10e+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.7.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

0:0.08

1:0.08

2:0.11

3:0.06

4:0.06

5:0.07

6:0.09

7:0.12

8:0.06

9:0.07

10:0.07

11:0.08

Negative Logits

reason

-2.34

externalActionCode

-1.96

 Perception

-1.95

 answ

-1.92

 ignore

-1.89

 ancest

-1.86

rs

-1.78

Reason

-1.78

know

-1.77

 impossibility

-1.76

POSITIVE LOGITS

imentary

2.56

Thumbnail

2.24

ridor

1.98

neau

1.87

 Ambro

1.86

raltar

1.84

glomer

1.83

gel

1.83

 dinners

1.75

 iPads

1.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.