INDEX

Explanations

phrases related to competition or conflict

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

phrases related to actions, particularly those expressing intention or direction

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

vette

-0.69

PLA

-0.68

 Mobil

-0.67

rov

-0.66

Riv

-0.65

iven

-0.64

iev

-0.64

 intervene

-0.64

oti

-0.63

 improv

-0.62

POSITIVE LOGITS

 Head

2.22

 Heads

2.21

Head

2.18

 head

2.10

 heads

2.08

head

1.95

 HEAD

1.94

heads

1.92

HEAD

1.74

 tails

1.61

Activations Density 0.170%

phrases related to competition or conflict

phrases related to actions, particularly those expressing intention or direction

No Comments

No Known Activations