Upload best model and additional files

adapter_config.json

@@ -23,12 +23,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj",
-    "o_proj",
     "k_proj",
-    "down_proj",
     "v_proj",
     "gate_proj",
+    "q_proj",
+    "o_proj",
+    "down_proj",
     "up_proj"
   ],
   "task_type": "CAUSAL_LM",

evaluation_results.json

@@ -0,0 +1,70 @@
+{
+    "1": {
+        "lora_alpha": 23,
+        "lora_dropout": 0.25385249450462455,
+        "learning_rate": 0.0002521627789110728,
+        "gradient_accumulation_steps": 3,
+        "eval_results": {
+            "meteor_scores": {
+                "meteor": 0.0024752475247524753
+            },
+            "rouge_scores": {
+                "rouge1": 0.0,
+                "rouge2": 0.0,
+                "rougeL": 0.0,
+                "rougeLsum": 0.0
+            },
+            "bleu_scores": {
+                "bleu": 0.0,
+                "precisions": [
+                    0.014184397163120567,
+                    0.0,
+                    0.0,
+                    0.0
+                ],
+                "brevity_penalty": 0.6173806530665256,
+                "length_ratio": 0.6746411483253588,
+                "translation_length": 141,
+                "reference_length": 209
+            },
+            "perplexity": 192550048.0
+        }
+    },
+    "2": {
+        "lora_alpha": 30,
+        "lora_dropout": 0.2587925120959257,
+        "learning_rate": 0.0001665618077288838,
+        "gradient_accumulation_steps": 2,
+        "eval_results": {
+            "meteor_scores": {
+                "meteor": 0.017326732673267325
+            },
+            "rouge_scores": {
+                "rouge1": 0.0,
+                "rouge2": 0.0,
+                "rougeL": 0.0,
+                "rougeLsum": 0.0
+            },
+            "bleu_scores": {
+                "bleu": 0.0,
+                "precisions": [
+                    0.05673758865248227,
+                    0.0,
+                    0.0,
+                    0.0
+                ],
+                "brevity_penalty": 0.6173806530665256,
+                "length_ratio": 0.6746411483253588,
+                "translation_length": 141,
+                "reference_length": 209
+            },
+            "perplexity": 139593840.0
+        }
+    },
+    "best_param": {
+        "lora_alpha": 30.41238686657094,
+        "lora_dropout": 0.2587925120959257,
+        "learning_rate": 0.0001665618077288838,
+        "gradient_accumulation_steps": 2
+    }
+}

lora_finetuning.log

@@ -0,0 +1,66 @@
+2025-01-28 12:10:31,187 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 12:20:17,720 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 13:10:47,232 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 13:10:50,924 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 13:15:32,374 - Using default tokenizer.
+2025-01-28 13:15:34,813 - Evaluation Results: {'meteor_scores': {'meteor': 0.0024752475247524753}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.014184397163120567, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 212501856.0}
+2025-01-28 13:15:34,846 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 13:15:34,885 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 13:24:59,597 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 13:29:13,244 - Using default tokenizer.
+2025-01-28 13:29:16,262 - Evaluation Results: {'meteor_scores': {'meteor': 0.017326732673267325}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.05673758865248227, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 133829632.0}
+2025-01-28 13:29:16,268 - Best hyperparameters: {'gradient_accumulation_steps': 2.2367355022465674, 'learning_rate': 0.0001665618077288838, 'lora_alpha': 30.41238686657094, 'lora_dropout': 0.2587925120959257}
+2025-01-28 13:29:16,289 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:12:50,446 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 14:12:53,816 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:15:36,548 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 14:15:40,086 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:20:21,304 - Using default tokenizer.
+2025-01-28 14:20:23,229 - Evaluation Results: {'meteor_scores': {'meteor': 0.0024752475247524753}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.014184397163120567, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 184389056.0}
+2025-01-28 14:20:23,256 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 14:20:23,279 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:20:26,700 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:24:39,399 - Using default tokenizer.
+2025-01-28 14:24:41,483 - Evaluation Results: {'meteor_scores': {'meteor': 0.017326732673267325}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.05673758865248227, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 136236672.0}
+2025-01-28 14:24:41,489 - Best hyperparameters: {'gradient_accumulation_steps': 2.2367355022465674, 'learning_rate': 0.0001665618077288838, 'lora_alpha': 30.41238686657094, 'lora_dropout': 0.2587925120959257}
+2025-01-28 14:24:41,512 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:30:38,720 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 14:30:42,775 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:35:24,928 - Using default tokenizer.
+2025-01-28 14:35:26,918 - Evaluation Results: {'meteor_scores': {'meteor': 0.039603960396039604}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.11347517730496454, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 194999696.0}
+2025-01-28 14:35:26,944 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 14:35:26,974 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:35:30,530 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:39:42,678 - Using default tokenizer.
+2025-01-28 14:39:44,703 - Evaluation Results: {'meteor_scores': {'meteor': 0.017326732673267325}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.05673758865248227, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 140277120.0}
+2025-01-28 14:39:44,707 - Best hyperparameters: {'gradient_accumulation_steps': 2.2367355022465674, 'learning_rate': 0.0001665618077288838, 'lora_alpha': 30.41238686657094, 'lora_dropout': 0.2587925120959257}
+2025-01-28 14:39:44,728 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:51:34,496 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 14:51:38,389 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 14:56:21,131 - Using default tokenizer.
+2025-01-28 14:56:23,242 - Evaluation Results: {'meteor_scores': {'meteor': 0.0024752475247524753}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.014184397163120567, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 201304048.0}
+2025-01-28 14:56:23,271 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 14:56:23,363 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 14:56:26,765 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 15:00:41,363 - Using default tokenizer.
+2025-01-28 15:00:43,468 - Evaluation Results: {'meteor_scores': {'meteor': 0.017326732673267325}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.05673758865248227, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 130403984.0}
+2025-01-28 15:00:43,472 - Best hyperparameters: {'gradient_accumulation_steps': 2.2367355022465674, 'learning_rate': 0.0001665618077288838, 'lora_alpha': 30.41238686657094, 'lora_dropout': 0.2587925120959257}
+2025-01-28 15:00:43,493 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 15:02:25,610 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 15:02:29,454 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 15:07:10,155 - Using default tokenizer.
+2025-01-28 15:07:12,488 - Evaluation Results: {'meteor_scores': {'meteor': 0.0024752475247524753}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.014184397163120567, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 211998656.0}
+2025-01-28 15:07:12,571 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 15:07:12,672 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 15:07:16,934 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 15:12:17,773 - Training with params: lora_alpha=23, lora_dropout=0.25385249450462455, learning_rate=0.0002521627789110728, gradient_accumulation_steps=3
+2025-01-28 15:12:21,432 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 15:17:01,601 - Using default tokenizer.
+2025-01-28 15:17:03,730 - Evaluation Results: {'meteor_scores': {'meteor': 0.0024752475247524753}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.014184397163120567, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 192550048.0}
+2025-01-28 15:17:03,763 - Training with params: lora_alpha=30, lora_dropout=0.2587925120959257, learning_rate=0.0001665618077288838, gradient_accumulation_steps=2
+2025-01-28 15:17:03,790 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!
+2025-01-28 15:17:07,194 - Detected kernel version 5.4.0, which is below the recommended minimum of 5.5.0; this can cause the process to hang. It is recommended to upgrade the kernel to the minimum version or higher.
+2025-01-28 15:21:19,175 - Using default tokenizer.
+2025-01-28 15:21:21,237 - Evaluation Results: {'meteor_scores': {'meteor': 0.017326732673267325}, 'rouge_scores': {'rouge1': 0.0, 'rouge2': 0.0, 'rougeL': 0.0, 'rougeLsum': 0.0}, 'bleu_scores': {'bleu': 0.0, 'precisions': [0.05673758865248227, 0.0, 0.0, 0.0], 'brevity_penalty': 0.6173806530665256, 'length_ratio': 0.6746411483253588, 'translation_length': 141, 'reference_length': 209}, 'perplexity': 139593840.0}
+2025-01-28 15:21:21,243 - Best hyperparameters: {'gradient_accumulation_steps': 2.2367355022465674, 'learning_rate': 0.0001665618077288838, 'lora_alpha': 30.41238686657094, 'lora_dropout': 0.2587925120959257}
+2025-01-28 15:21:21,264 - Already found a `peft_config` attribute in the model. This will lead to having multiple adapters in the model. Make sure to know what you are doing!