Upload folder using huggingface_hub

.gitattributes

@@ -38,3 +38,5 @@ part1/moe-shared/routing-heatmap.png filter=lfs diff=lfs merge=lfs -text
 part1/moe-shared/translations.jsonl filter=lfs diff=lfs merge=lfs -text
 part1/moe-top1/routing-heatmap.png filter=lfs diff=lfs merge=lfs -text
 part1/moe-top1/translations.jsonl filter=lfs diff=lfs merge=lfs -text
+part1/moe-top2/routing-heatmap.png filter=lfs diff=lfs merge=lfs -text
+part1/moe-top2/translations.jsonl filter=lfs diff=lfs merge=lfs -text

part1/moe-top2/config.json

@@ -0,0 +1,62 @@
+{
+  "model": {
+    "d_model": 512,
+    "n_layers": 6,
+    "n_heads": 8,
+    "n_ctx": 256,
+    "d_ff": 1536,
+    "dropout": 0.0,
+    "use_moe": true,
+    "n_experts": 4,
+    "top_k": 2,
+    "n_shared": 0,
+    "d_expert": 384
+  },
+  "seed": 0,
+  "data": {
+    "name": "belumind/en-vi-ja-curated-500k-triplets",
+    "splits": {
+      "train": "train",
+      "val": "validation",
+      "test": "test"
+    },
+    "columns": {
+      "en": "en",
+      "vi": "vi",
+      "ja": "ja"
+    },
+    "seed": 0,
+    "limit": null
+  },
+  "tokenizer": {
+    "file": "unigram-16k.json",
+    "vocab_size": 16000,
+    "sample_triplets": 200000
+  },
+  "train": {
+    "max_tokens": 60000000,
+    "batch_size": 128,
+    "lr": 0.001,
+    "betas": [
+      0.9,
+      0.98
+    ],
+    "weight_decay": 0.1,
+    "grad_clip": 1.0,
+    "warmup_fraction": 0.05,
+    "min_lr_ratio": 0.1,
+    "eval_every_fraction": 0.1,
+    "log_every": 50
+  },
+  "evaluate": {
+    "batch_size": 128,
+    "val_pairs": 4000,
+    "max_new_tokens": 128,
+    "bleu_limit": null
+  },
+  "wandb": {
+    "enabled": true,
+    "project": "anlp-assignment-2"
+  },
+  "name": "moe-top2"
+}

part1/moe-top2/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:254dd1330f78254f2eb2bd82533948c27085f9fbe683703048fcb4deeca3cb56
+size 147439711

part1/moe-top2/results.json

@@ -0,0 +1,14 @@
+{
+  "name": "moe-top2",
+  "parameters": {
+    "total": 36850176,
+    "active": 29772288
+  },
+  "complete": true,
+  "test_loss": 1.3488332055230736,
+  "test_perplexity": 3.8529273326392515,
+  "bleu_vi": 42.716839237328806,
+  "bleu_ja": 34.223923777649865,
+  "bleu": 38.56568762958229,
+  "bleu_pairs": 49583
+}

part1/moe-top2/translations.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa204af835b2b7ae169ada4a7a8936a2e27da0fe4a2cf2faf47f104d36b9c9eb
+size 16628131