INDEX

Explanations

conversations between different individuals, possibly with some conflict or disagreement

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

elements of dialogue, particularly interjections and speaker labels in conversations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.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

1:0.09

2:0.08

3:0.09

4:0.04

5:0.22

6:0.06

7:0.02

8:0.07

9:0.09

10:0.09

11:0.03

Negative Logits

MFT

-1.38

 gradient

-1.38

versions

-1.28

 Mexicans

-1.27

Grad

-1.26

 Mexican

-1.24

cro

-1.24

 suitable

-1.23

 sensitive

-1.20

alist

-1.20

POSITIVE LOGITS

answer

1.75

lein

1.50

 Answer

1.50

utch

1.46

kie

1.44

rike

1.44

leen

1.44

clair

1.43

 Answers

1.43

 Puzzle

1.41

Activations Density 0.012%

conversations between different individuals, possibly with some conflict or disagreement

elements of dialogue, particularly interjections and speaker labels in conversations

No Comments

No Known Activations