INDEX

Explanations

words related to languages or linguistic activities

oai_token-act-pair · gpt-3.5-turbo

repeated references to "language."

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

destroy

-0.67

ãģ®éŃĶ

-0.67

iatus

-0.66

bryce

-0.66

 Presents

-0.66

aunt

-0.65

invest

-0.64

Ul

-0.61

pless

-0.61

 Triumph

-0.61

POSITIVE LOGITS

 language

3.84

 Language

3.02

language

2.57

 languages

2.54

Language

2.51

 Languages

2.19

anguage

2.04

 linguistic

1.84

anguages

1.82

 vocabulary

1.77

Activations Density 0.021%

words related to languages or linguistic activities

repeated references to "language."

No Comments

No Known Activations