INDEX

Explanations

phrases related to contradictions or negations in context

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

0:0.03

1:0.02

2:0.41

3:0.06

4:0.06

5:0.03

6:0.13

7:0.02

8:0.04

9:0.03

10:0.05

11:0.05

Negative Logits

 Gael

-1.99

iHUD

-1.73

itus

-1.57

 Kills

-1.53

ogl

-1.51

 qualification

-1.50

 Rain

-1.49

ビ

-1.49

 proportions

-1.43

Scope

-1.43

POSITIVE LOGITS

 themselves

2.32

selves

2.21

 selves

1.98

ently

1.75

 THEIR

1.74

undai

1.70

okers

1.70

itimate

1.66

tten

1.63

 abundantly

1.62

Activations Density 0.055%

phrases related to contradictions or negations in context

No Comments

No Known Activations