INDEX

Explanations

locations or pathways like "corridors"

oai_token-act-pair · gpt-3.5-turbo

references to various types of corridors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

dom

-0.71

iak

-0.71

mol

-0.71

ilities

-0.68

olute

-0.68

odium

-0.68

omi

-0.68

DEF

-0.66

ken

-0.65

held

-0.65

POSITIVE LOGITS

 Corridor

1.36

 corridors

1.34

 corridor

1.32

ridor

1.13

ttes

0.84

 routes

0.82

 route

0.81

 passages

0.80

 alley

0.77

 ventilation

0.76

Activations Density 0.007%

locations or pathways like "corridors"

references to various types of corridors

No Comments

No Known Activations