INDEX

Explanations

No Explanations Found

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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.06

2:0.07

3:0.09

4:0.09

5:0.06

6:0.08

7:0.09

8:0.09

9:0.08

10:0.08

11:0.08

Negative Logits

conservancy

-2.09

escal

-2.07

combe

-1.90

organized

-1.89

oided

-1.84

vet

-1.79

vironment

-1.78

irs

-1.77

 stockpile

-1.75

igslist

-1.72

POSITIVE LOGITS

}}}

2.08

版

2.03

��

1.90

�

1.83

Ax

1.76

 Miche

1.67

Peg

1.67

��

1.67

HW

1.66

 Param

1.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.