INDEX

Explanations

instances where the word "look" appears, with a focus on phrases that involve observation, examination, or assessment

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

orb

-0.63

ä¹

-0.61

CENT

-0.60

Cele

-0.58

Osw

-0.57

rament

-0.56

 restraining

-0.56

learning

-0.56

MQ

-0.55

theless

-0.55

POSITIVE LOGITS

ahead

0.97

 suspic

0.82

ãĤ¶

0.74

 favorably

0.74

 alike

0.73

 forward

0.72

ared

0.69

 into

0.67

attered

0.66

isite

0.66

Activations Density 0.400%

instances where the word "look" appears, with a focus on phrases that involve observation, examination, or assessment

No Comments

No Known Activations