INDEX

Explanations

phrases related to negative consequences or challenges

oai_token-act-pair · gpt-3.5-turbo

concluding statements or phrases that summarize a point or thought

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 deity

-0.77

 brief

-0.75

 compliment

-0.72

nesday

-0.70

clus

-0.70

ilion

-0.69

iphate

-0.69

 medium

-0.69

 breeze

-0.68

rera

-0.68

POSITIVE LOGITS

 Worse

1.84

 Fortunately

1.42

 Unable

1.40

 Consequently

1.35

 Luckily

1.35

 Thankfully

1.30

 Moreover

1.21

 Meanwhile

1.19

 Eventually

1.15

 Symptoms

1.15

Activations Density 0.652%

phrases related to negative consequences or challenges

concluding statements or phrases that summarize a point or thought

No Comments

No Known Activations