INDEX

Explanations

references to sports, with a focus on teams losing

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lost" related to game outcomes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ahu

-0.74

hire

-0.67

Companies

-0.65

kson

-0.64

ographies

-0.62

POL

-0.60

SPONSORED

-0.59

 rods

-0.59

IPS

-0.58

NVIDIA

-0.57

POSITIVE LOGITS

 miser

1.11

 convinc

1.02

 decisively

0.90

 bitterly

0.79

 horribly

0.79

to

0.78

 badly

0.78

 narrowly

0.74

 heartbreaking

0.72

 midway

0.69

Activations Density 0.096%

references to sports, with a focus on teams losing

instances of the word "lost" related to game outcomes

No Comments

No Known Activations