INDEX

Explanations

the word "soft"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "soft" and its variants, likely in contexts that relate to softness or gentleness

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

 rotting

-0.71

 bananas

-0.70

 hunger

-0.70

 graveyard

-0.69

 ticking

-0.69

py

-0.68

 Timbers

-0.67

dru

-0.67

 pollen

-0.64

 cocaine

-0.64

POSITIVE LOGITS

soft

3.54

Soft

2.98

 Soft

2.79

 Interactive

2.06

Games

1.24

 Software

1.22

Software

1.19

 Dynamics

1.15

oft

1.14

 Computing

1.07

Activations Density 0.026%

the word "soft"

occurrences of the word "soft" and its variants, likely in contexts that relate to softness or gentleness

No Comments

No Known Activations