add plots

Samoed · May 13, 2024 · a7f4cb0 · a7f4cb0
1 parent 990b69c
commit a7f4cb0
Show file tree

Hide file tree

Showing 40 changed files with 22,856 additions and 9 deletions.
diff --git a/notebooks/plots.ipynb b/notebooks/plots.ipynb
diff --git a/notebooks/plots/bert.png b/notebooks/plots/bert.png
diff --git a/notebooks/plots/bert.svg b/notebooks/plots/bert.svg
diff --git a/notebooks/plots/bert_Llama-3.png b/notebooks/plots/bert_Llama-3.png
diff --git a/notebooks/plots/bert_Llama-3.svg b/notebooks/plots/bert_Llama-3.svg
diff --git a/notebooks/plots/bert_Phi-3.png b/notebooks/plots/bert_Phi-3.png
diff --git a/notebooks/plots/bert_Phi-3.svg b/notebooks/plots/bert_Phi-3.svg
diff --git a/notebooks/plots/chrf.png b/notebooks/plots/chrf.png
diff --git a/notebooks/plots/chrf.svg b/notebooks/plots/chrf.svg
diff --git a/notebooks/plots/eval_bert_f1.png b/notebooks/plots/eval_bert_f1.png
diff --git a/notebooks/plots/eval_bert_f1.svg b/notebooks/plots/eval_bert_f1.svg
diff --git a/notebooks/plots/eval_bert_metrics.png b/notebooks/plots/eval_bert_metrics.png
diff --git a/notebooks/plots/eval_bert_metrics.svg b/notebooks/plots/eval_bert_metrics.svg
diff --git a/notebooks/plots/eval_bert_precision.png b/notebooks/plots/eval_bert_precision.png
diff --git a/notebooks/plots/eval_bert_precision.svg b/notebooks/plots/eval_bert_precision.svg
diff --git a/notebooks/plots/eval_bert_recall.png b/notebooks/plots/eval_bert_recall.png
diff --git a/notebooks/plots/eval_bert_recall.svg b/notebooks/plots/eval_bert_recall.svg
diff --git a/notebooks/plots/eval_chrf_score.png b/notebooks/plots/eval_chrf_score.png
diff --git a/notebooks/plots/eval_chrf_score.svg b/notebooks/plots/eval_chrf_score.svg
diff --git a/notebooks/plots/eval_rouge1.png b/notebooks/plots/eval_rouge1.png
diff --git a/notebooks/plots/eval_rouge1.svg b/notebooks/plots/eval_rouge1.svg
diff --git a/notebooks/plots/eval_rouge2.png b/notebooks/plots/eval_rouge2.png
diff --git a/notebooks/plots/eval_rouge2.svg b/notebooks/plots/eval_rouge2.svg
diff --git a/notebooks/plots/eval_rougeL.png b/notebooks/plots/eval_rougeL.png
diff --git a/notebooks/plots/eval_rougeL.svg b/notebooks/plots/eval_rougeL.svg
diff --git a/notebooks/plots/eval_rougeLsum.png b/notebooks/plots/eval_rougeLsum.png
diff --git a/notebooks/plots/eval_rougeLsum.svg b/notebooks/plots/eval_rougeLsum.svg
diff --git a/notebooks/plots/eval_rouge_metrics.png b/notebooks/plots/eval_rouge_metrics.png
diff --git a/notebooks/plots/eval_rouge_metrics.svg b/notebooks/plots/eval_rouge_metrics.svg
diff --git a/notebooks/plots/eval_score.png b/notebooks/plots/eval_score.png
diff --git a/notebooks/plots/eval_score.svg b/notebooks/plots/eval_score.svg
diff --git a/notebooks/plots/loss.png b/notebooks/plots/loss.png
diff --git a/notebooks/plots/loss.svg b/notebooks/plots/loss.svg
diff --git a/notebooks/plots/rouge.png b/notebooks/plots/rouge.png
diff --git a/notebooks/plots/rouge.svg b/notebooks/plots/rouge.svg
diff --git a/notebooks/plots/rouge_Llama-3.png b/notebooks/plots/rouge_Llama-3.png
diff --git a/notebooks/plots/rouge_Llama-3.svg b/notebooks/plots/rouge_Llama-3.svg
diff --git a/notebooks/plots/rouge_Phi-3.png b/notebooks/plots/rouge_Phi-3.png
diff --git a/notebooks/plots/rouge_Phi-3.svg b/notebooks/plots/rouge_Phi-3.svg
diff --git a/notebooks/trainunsloth.ipynb b/notebooks/trainunsloth.ipynb
@@ -802,9 +802,7 @@
     "    labels = np.where(labels != -100, labels, tokenizer.pad_token_id)\n",
     "    labels = tokenizer.batch_decode(labels, skip_special_tokens=True)\n",
     "    rouge_score = rouge.compute(predictions=predictions, references=labels)\n",
-    "    bert_score = bertscore.compute(\n",
-    "        predictions=predictions, references=[[label] for label in labels], lang=\"en\"\n",
-    "    )\n",
+    "    bert_score = bertscore.compute(predictions=predictions, references=[[label] for label in labels], lang=\"en\")\n",
     "\n",
     "    chrf_score = chrf.compute(predictions=predictions, references=labels, word_order=2)\n",
     "\n",
@@ -1077,9 +1075,7 @@
     "    output_texts = []\n",
     "    if isinstance(examples, dict):\n",
     "        return PROMPT.format(examples[\"prompt\"], examples[\"completion\"])\n",
-    "    for prompt, completion in zip(\n",
-    "        examples[\"prompt\"], examples[\"completion\"], strict=False\n",
-    "    ):\n",
+    "    for prompt, completion in zip(examples[\"prompt\"], examples[\"completion\"], strict=False):\n",
     "        text = PROMPT.format(prompt, completion)\n",
     "        output_texts.append(text)\n",
     "    return output_texts"
@@ -1539,9 +1535,7 @@
     "used_percentage = round(used_memory / max_memory * 100, 3)\n",
     "lora_percentage = round(used_memory_for_lora / max_memory * 100, 3)\n",
     "print(f\"{trainer_stats.metrics['train_runtime']} seconds used for training.\")\n",
-    "print(\n",
-    "    f\"{round(trainer_stats.metrics['train_runtime']/60, 2)} minutes used for training.\"\n",
-    ")\n",
+    "print(f\"{round(trainer_stats.metrics['train_runtime']/60, 2)} minutes used for training.\")\n",
     "print(f\"Peak reserved memory = {used_memory} GB.\")\n",
     "print(f\"Peak reserved memory for training = {used_memory_for_lora} GB.\")\n",
     "print(f\"Peak reserved memory % of max memory = {used_percentage} %.\")\n",