INDEX

Explanations

phrases that indicate actions or attempts to perform tasks

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_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.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

0:0.01

1:0.02

2:0.16

3:0.16

4:0.01

5:0.02

6:0.07

7:0.08

8:0.04

9:0.20

10:0.09

11:0.08

Negative Logits

 Travels

-1.13

workers

-1.01

 Kinnikuman

-0.98

 Chronicles

-0.91

 Carbuncle

-0.91

 Centers

-0.90

houses

-0.90

rises

-0.90

 Belief

-0.90

akings

-0.89

POSITIVE LOGITS

��

1.11

pless

1.10

anut

1.08

ocre

1.07

ipal

1.06

 cannabin

1.04

 unwitting

1.01

leck

1.00

acea

0.99

 recre

0.98

Activations Density 0.051%

phrases that indicate actions or attempts to perform tasks

No Comments

No Known Activations