INDEX

Explanations

terms related to geographical locations or landmarks

oai_token-act-pair · gpt-3.5-turbo

the word "est" in various contexts, likely indicating a focus on comparative or superlative forms

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

 Zombies

-0.78

 Chandra

-0.72

 Whip

-0.71

 Graham

-0.70

 Spike

-0.67

 Bullets

-0.67

 Benny

-0.66

sbm

-0.66

 Planes

-0.66

 Plane

-0.66

POSITIVE LOGITS

est

3.99

 estrogen

1.13

mi

1.08

 este

1.05

PDT

1.01

 gulf

1.00

es

1.00

 mascul

0.98

EST

0.97

lag

0.94

Activations Density 0.006%

terms related to geographical locations or landmarks

the word "est" in various contexts, likely indicating a focus on comparative or superlative forms

No Comments

No Known Activations