INDEX

Explanations

words related to names, particularly those ending in "nes"

oai_token-act-pair · gpt-3.5-turbo

words related to "Nes" or "Nes" as a recurring motif within varied contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

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

 tangled

-0.66

JFK

-0.64

 McGregor

-0.64

 capital

-0.63

 spiked

-0.62

 poop

-0.62

big

-0.59

 cropped

-0.58

TR

-0.57

 steep

-0.57

POSITIVE LOGITS

nes

4.67

NES

2.23

nesia

1.67

nis

1.66

ne

1.64

nia

1.40

na

1.40

nie

1.32

ns

1.30

nos

1.28

Activations Density 0.006%

words related to names, particularly those ending in "nes"

words related to "Nes" or "Nes" as a recurring motif within varied contexts

No Comments

No Known Activations