INDEX

Explanations

proper nouns related to technology or titles

oai_token-act-pair · gpt-3.5-turbo

references to 'per' or its variations in numerical 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

seq

-0.60

 trenches

-0.60

 Equality

-0.58

 Hook

-0.57

jaw

-0.57

HQ

-0.57

 circles

-0.56

 graft

-0.56

IG

-0.56

Jem

-0.55

POSITIVE LOGITS

per

4.25

pering

2.12

PER

2.09

pered

1.96

pers

1.84

Per

1.79

perm

1.73

perate

1.65

Per

1.45

pert

1.43

Activations Density 0.016%

proper nouns related to technology or titles

references to 'per' or its variations in numerical contexts

No Comments

No Known Activations