INDEX

Explanations

mentions of negative outcomes, specifically losses

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "loss" in various 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Instruct

-0.67

 indo

-0.66

 Surve

-0.66

omet

-0.65

imaru

-0.65

commun

-0.64

Instruct

-0.63

JB

-0.63

ç«

-0.62

ulhu

-0.62

POSITIVE LOGITS

 loss

3.76

 Loss

2.94

loss

2.91

 losses

2.62

 losing

1.74

 lose

1.60

 lost

1.55

 defeat

1.46

 loses

1.42

 setback

1.40

Activations Density 0.019%

mentions of negative outcomes, specifically losses

occurrences of the word "loss" in various contexts

No Comments

No Known Activations