INDEX

Explanations

words related to specific locations or organizations, particularly ones with "ab" in them

oai_token-act-pair · gpt-3.5-turbo

occurrences of the substring "ab"

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 arrivals

-0.68

 Tycoon

-0.66

 Sovereign

-0.66

 orders

-0.65

 additions

-0.64

 order

-0.63

 scen

-0.63

terday

-0.62

 belts

-0.61

 numbered

-0.60

POSITIVE LOGITS

ab

4.01

abs

2.30

abo

2.09

abus

2.04

abe

2.04

aban

1.85

abi

1.84

abee

1.81

aba

1.80

abal

1.77

Activations Density 0.020%

words related to specific locations or organizations, particularly ones with "ab" in them

occurrences of the substring "ab"

No Comments

No Known Activations