Upload folder using huggingface_hub

.gitattributes

@@ -40,3 +40,5 @@ 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-matched/routing-heatmap.png filter=lfs diff=lfs merge=lfs -text
+part1/moe-top2-matched/translations.jsonl filter=lfs diff=lfs merge=lfs -text

part1/moe-top2-matched/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": 768
+  },
+  "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-matched"
+}

part1/moe-top2-matched/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:830b77196248fc9e4dd9e996897f80abf16c35e77d2f771bba8cb34d96db0aff
+size 204062815

part1/moe-top2-matched/results.json

@@ -0,0 +1,14 @@
+{
+  "name": "moe-top2-matched",
+  "parameters": {
+    "total": 51005952,
+    "active": 36850176
+  },
+  "complete": true,
+  "test_loss": 1.3337380854112106,
+  "test_perplexity": 3.795203700435645,
+  "bleu_vi": 43.04103245629448,
+  "bleu_ja": 34.59553595492863,
+  "bleu": 38.84703560242226,
+  "bleu_pairs": 49583
+}

part1/moe-top2-matched/translations.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cadfb315ff0aea784ac307e8c4185c2895e6c90d1f319f5d0b2a0304cb475279
+size 16621112