zad9 bow

2023-07-02 12:47:23 +02:00 · 2023-07-02 12:47:23 +02:00 · 4173fd8cd3
commit 4173fd8cd3
18 changed files with 1048786 additions and 0 deletions
--- a/README.md
+++ b/README.md
@ -0,0 +1,9 @@
+Challenging America word-gap prediction
+===================================
+
+Guess a word in a gap.
+
+Evaluation metric
+-----------------
+
+LikelihoodHashed is the metric
--- a/config.txt
+++ b/config.txt
@ -0,0 +1 @@
+--metric PerplexityHashed --precision 2  --in-header in-header.tsv  --out-header out-header.tsv
--- a/config/params.yaml
+++ b/config/params.yaml
@ -0,0 +1,9 @@
+epochs: 3
+embed_size: 256
+device: cuda
+vocab_size: 35000
+batch_size: 3200
+learning_rate: 0.001
+k: 15
+wildcard_minweight: 0.01
+beta: 0.85
--- a/dev-0/expected.tsv
+++ b/dev-0/expected.tsv
--- a/dev-0/hate-speech-info.tsv
+++ b/dev-0/hate-speech-info.tsv
--- a/dev-0/in.tsv.xz
+++ b/dev-0/in.tsv.xz
--- a/dev-0/out.tsv
+++ b/dev-0/out.tsv
--- a/filename.pickle
+++ b/filename.pickle
--- a/gonito.yaml
+++ b/gonito.yaml
@ -0,0 +1,14 @@
+description: zad8, trigram with left/right context embeddings
+tags:
+  - neural-network
+  - left-to-right
+params:
+  epochs: 3
+  learning-rate: 0.0003
+  vocab-size: 20000
+  batch_s: 3200
+  top_k_words: 20
+param-files:
+  - config/*.yaml
+links:
+  - repo: "https://git.wmi.amu.edu.pl/s470618/challenging-america-word-gap-prediction"
--- a/in-header.tsv
+++ b/in-header.tsv
@ -0,0 +1 @@
+FileId	Year	LeftContext	RightContext
--- a/model-hexa-bag-following-bs3200-lr0.001embed256.bin
+++ b/model-hexa-bag-following-bs3200-lr0.001embed256.bin
--- a/out-header.tsv
+++ b/out-header.tsv
@ -0,0 +1 @@
+Word
--- a/test-A/hate-speech-info.tsv
+++ b/test-A/hate-speech-info.tsv
--- a/test-A/in.tsv.xz
+++ b/test-A/in.tsv.xz
--- a/test-A/out.tsv
+++ b/test-A/out.tsv
--- a/train/expected.tsv
+++ b/train/expected.tsv
--- a/train/hate-speech-info.tsv
+++ b/train/hate-speech-info.tsv
--- a/zad9.ipynb
+++ b/zad9.ipynb
				`@ -0,0 +1 @@`
				`--metric PerplexityHashed --precision 2 --in-header in-header.tsv --out-header out-header.tsv`