INDEX

Explanations

terms that include a particular ending or sequence of characters, specifically 'fo' and 'fu'

oai_token-act-pair · gpt-3.5-turbo

specific references to food-related topics

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

 reciproc

-0.77

 unpaid

-0.72

 intraven

-0.66

 liability

-0.65

 goodwill

-0.64

 Galile

-0.62

URES

-0.61

Nir

-0.61

ERSON

-0.61

 Singer

-0.60

POSITIVE LOGITS

fo

3.97

fu

2.15

1.79

fc

1.70

fs

1.68

fb

1.58

fa

1.58

fac

1.57

fing

1.54

fen

1.51

Activations Density 0.025%

terms that include a particular ending or sequence of characters, specifically 'fo' and 'fu'

specific references to food-related topics

No Comments

No Known Activations