tor-simt-llama-3-8b-curated-en-vi — git@unknown

.gitattributes

@@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+annotate/ar-en/matrices.jsonl filter=lfs diff=lfs merge=lfs -text
+annotate/de-en/matrices.jsonl filter=lfs diff=lfs merge=lfs -text
+annotate/en-ar/matrices.jsonl filter=lfs diff=lfs merge=lfs -text
+annotate/en-de/matrices.jsonl filter=lfs diff=lfs merge=lfs -text
+annotate/en-ru/matrices.jsonl filter=lfs diff=lfs merge=lfs -text
+annotate/en-vi/matrices.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,45 @@
+---
+tags:
+- simultaneous-translation
+- east
+- ot-annotation
+language: ['ar', 'de', 'en', 'ru', 'vi']
+license: mit
+---
+
+# llama_3_8b_curated
+
+Experiment `llama_3_8b_curated` from the *Teacher-Free Read/Write Annotation for Simultaneous
+Machine Translation* project.
+
+## Recipe
+
+- **Backbone:** `meta-llama/Meta-Llama-3-8B-Instruct`
+- **Corpus:** `curated`
+- **Annotator:** `same_as_backbone`
+- **Criterion:** `ot` (τ = 0.3)
+- **Latencies:** ['low', 'medium', 'high']
+
+## Files in this repo
+
+- `config.yaml` — the exact experiment config that produced this run.
+- `manifest.json` — git sha, hostname, GPUs, timestamps.
+- `logs/` — per-stage stdout+stderr from `bin/run`.
+- `eval/` — every landed eval-JSON cell (hypothesis, reference, AL, BLEU).
+- `annotate/` — per-direction `matrices.jsonl` (divergence matrices) produced
+  by this backbone as annotator.
+- `source_pool.json` — the corpus rows the matrices index into (`{index, source,
+  target, src_lang, tgt_lang, latency, source_chunks, target_chunks, _corpus}`).
+  Matrices are unjoinable without this file — records only carry `index`.
+- SFT checkpoint (`*.safetensors` + tokenizer).
+
+## Reproduce
+
+```bash
+git clone https://github.com/dipankarsrirag/simt-tor-26.git
+cd simt-tor-26
+cp .simtrc.example .simtrc  # edit paths for your setup
+bin/run configs/01_llama_3_8b_curated.yaml --ngpus N
+```
+
+Git commit at time of run: `unknown`

annotate/ar-en/DONE

@@ -0,0 +1 @@
+done

annotate/ar-en/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea9e9fb411e8844640fae96e8429c35e7c1a6f4a364563c0e37cb5a7b1d4766f
+size 146117990

annotate/ar-en/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 9891,
+    "n_sentences_annotated": 9891,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.9922151450813871,
+      "mean_committed_fraction": 0.9318136087865282,
+      "chunks_per_sentence_mean_ours": 5.663330300272976,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 9688,
+        "n_nan": 203,
+        "min": -4.950377051351219e-16,
+        "median": 0.9197230968323045,
+        "max": 1.0
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/de-en/DONE

@@ -0,0 +1 @@
+done

annotate/de-en/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afb47888d897f0e72e18e6101a93b09de8596a6fa17753ea1798af61ed55f96b
+size 227757323

annotate/de-en/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 9118,
+    "n_sentences_annotated": 9118,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.9964904584338671,
+      "mean_committed_fraction": 0.9775457410513922,
+      "chunks_per_sentence_mean_ours": 5.34218030269796,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 8902,
+        "n_nan": 216,
+        "min": -5.255955881681541e-16,
+        "median": 0.8871386018907612,
+        "max": 0.9919156224708092
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/en-ar/DONE

@@ -0,0 +1 @@
+done

annotate/en-ar/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d205d10707e0fc0e2ba630346d168bd5f7ea42aba93681e885c47a0581f50338
+size 173250818

annotate/en-ar/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 10000,
+    "n_sentences_annotated": 10000,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.9947,
+      "mean_committed_fraction": 0.9581987467828927,
+      "chunks_per_sentence_mean_ours": 6.221,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 9762,
+        "n_nan": 238,
+        "min": -6.873854856488296e-16,
+        "median": 0.9169409126628129,
+        "max": 1.0
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/en-de/DONE

@@ -0,0 +1 @@
+done

annotate/en-de/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74523acf2370777230c7f44369ecc8c66c86efe16ff3595fd459ad28af30c48e
+size 331552450

annotate/en-de/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 9936,
+    "n_sentences_annotated": 9936,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.9992954911433173,
+      "mean_committed_fraction": 0.995170410352499,
+      "chunks_per_sentence_mean_ours": 6.818941223832528,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 9807,
+        "n_nan": 129,
+        "min": -5.690079886834464e-16,
+        "median": 0.9225307913536488,
+        "max": 0.9983122664553511
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/en-ru/DONE

@@ -0,0 +1 @@
+done

annotate/en-ru/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:884fb1aea7b31e9c562e77f60757fd544589ca9ed7204c60d4bbf49baf98d62e
+size 254458426

annotate/en-ru/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 9979,
+    "n_sentences_annotated": 9979,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.9973945285098708,
+      "mean_committed_fraction": 0.9897734159045443,
+      "chunks_per_sentence_mean_ours": 7.079466880448943,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 9827,
+        "n_nan": 152,
+        "min": -4.950377051351219e-16,
+        "median": 0.9275602036706543,
+        "max": 0.9965765844070376
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/en-vi/DONE

@@ -0,0 +1 @@
+done

annotate/en-vi/matrices.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bd347efdee6574b3edea1648ee1077e75724ed00c94a91c9023be587c8053e0
+size 149167907

annotate/en-vi/summary.json

@@ -0,0 +1,40 @@
+{
+  "config": {
+    "n_sentences_requested": 50,
+    "n_sentences_kept": 9998,
+    "n_sentences_annotated": 9998,
+    "criterion": "ot",
+    "seed": 42,
+    "model_path": "/g/data/po67/dipankar/models/Meta-Llama-3-8B-Instruct",
+    "max_src_tokens": 80,
+    "taus": [
+      0.3
+    ],
+    "prompt_mode": "chat",
+    "record_entropy": false,
+    "indices_file": null,
+    "lookahead_k": 0
+  },
+  "sweep": [
+    {
+      "tau": 0.3,
+      "fire_fraction": 0.99249849969994,
+      "mean_committed_fraction": 0.9686583267700783,
+      "chunks_per_sentence_mean_ours": 6.039007801560312,
+      "chunks_per_sentence_mean_gpt4": 0.0,
+      "pearson_i_over_n_vs_j_over_m": {
+        "n_defined": 9689,
+        "n_nan": 309,
+        "min": -5.255955881681541e-16,
+        "median": 0.9258200997725514,
+        "max": 1.0
+      }
+    }
+  ],
+  "env": {
+    "device": "cuda",
+    "torch": "2.11.0+cu130",
+    "cuda_available": true,
+    "git_commit": "22c5889a5f88f1c8cd48728301457d04b01009d6"
+  }
+}

annotate/ru-en/DONE

@@ -0,0 +1 @@
+done