INDEX

Explanations

phrases indicating capability or sufficiency related to reading and comprehension

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

0:0.01

1:0.01

2:0.15

3:0.05

4:0.15

5:0.02

6:0.06

7:0.24

8:0.02

9:0.04

10:0.15

11:0.04

Negative Logits

 sweep

-1.58

Tro

-1.55

 rhet

-1.55

 annihilation

-1.49

campaign

-1.48

 POLIT

-1.45

direct

-1.45

ramid

-1.43

Rog

-1.42

alks

-1.42

POSITIVE LOGITS

remember

1.83

 reckoning

1.71

addon

1.66

cair

1.63

 stripes

1.62

 recol

1.59

aware

1.59

riter

1.56

comfort

1.55

<=

1.53

Activations Density 0.000%

phrases indicating capability or sufficiency related to reading and comprehension

No Comments

No Known Activations