loras commit

llama2_13b_responsibility-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-13b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "v_proj",
+    "o_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_13b_responsibility-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aadeb839f1fe30a28cbace1ac0f93a138b943a83f89c1aebe858375bbe157be6
+size 104973834

llama2_13b_responsibility-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,20 @@
+{
+  "base_model_name": "meta-llama_Llama-2-13b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.27035561203956604,
+  "eval_runtime": 318.5835,
+  "eval_samples_per_second": 4.85,
+  "eval_steps_per_second": 0.609,
+  "epoch": 4.0,
+  "current_steps": 6268,
+  "loss": 0.262,
+  "learning_rate": 1.5624999999999999e-06,
+  "train_runtime": 27128.8237,
+  "train_samples_per_second": 0.925,
+  "train_steps_per_second": 0.007,
+  "total_flos": 1.9848229987811328e+18,
+  "train_loss": 0.34869462008378943
+}

llama2_13b_responsibility-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_13b_responsibility-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 1024,
+  "dataset": "responsibility_train",
+  "eval_dataset": "responsibility_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_13b_responsibility-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}

llama2_13b_type-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-13b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "o_proj",
+    "v_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_13b_type-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:faac9479d9c6886ff2f0ff03b8a8e342ae3f3fe4346a71891a8cb545622caef5
+size 104973834

llama2_13b_type-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,20 @@
+{
+  "base_model_name": "meta-llama_Llama-2-13b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.6852580904960632,
+  "eval_runtime": 235.3491,
+  "eval_samples_per_second": 4.844,
+  "eval_steps_per_second": 0.608,
+  "epoch": 3.98,
+  "current_steps": 4586,
+  "loss": 0.6811,
+  "learning_rate": 8.571428571428571e-06,
+  "train_runtime": 17316.2264,
+  "train_samples_per_second": 1.07,
+  "train_steps_per_second": 0.008,
+  "total_flos": 1.4576995856705126e+18,
+  "train_loss": 0.7967557840877109
+}

llama2_13b_type-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_13b_type-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 1024,
+  "dataset": "type_train",
+  "eval_dataset": "type_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_13b_type-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}

llama2_70b_responsibility-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-70b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "o_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_70b_responsibility-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2911e088ce27e1bd3cabe145db8690212af86b05c580fba3639771db7894c296
+size 262376202

llama2_70b_responsibility-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,15 @@
+{
+  "base_model_name": "meta-llama_Llama-2-70b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.3329446017742157,
+  "eval_runtime": 797.899,
+  "eval_samples_per_second": 1.936,
+  "eval_steps_per_second": 0.243,
+  "epoch": 3.99,
+  "current_steps": 2174,
+  "loss": 0.3353,
+  "learning_rate": 0.00017202797202797203
+}

llama2_70b_responsibility-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_70b_responsibility-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 600,
+  "dataset": "responsibility_train",
+  "eval_dataset": "responsibility_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_70b_responsibility-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}

llama2_70b_type-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-70b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_70b_type-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:189dfae66be9aef6b175f6289d68d1534c1e8b301ebcf0d5987a52994fa69bde
+size 262376202

llama2_70b_type-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,20 @@
+{
+  "base_model_name": "meta-llama_Llama-2-70b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.5739039182662964,
+  "eval_runtime": 1244.9728,
+  "eval_samples_per_second": 0.916,
+  "eval_steps_per_second": 0.115,
+  "epoch": 3.99,
+  "current_steps": 2296,
+  "loss": 0.5677,
+  "learning_rate": 8.823529411764705e-06,
+  "train_runtime": 44904.3442,
+  "train_samples_per_second": 0.206,
+  "train_steps_per_second": 0.002,
+  "total_flos": 2.2812138370400256e+18,
+  "train_loss": 0.7556745294067595
+}

llama2_70b_type-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_70b_type-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 600,
+  "dataset": "type_train",
+  "eval_dataset": "type_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_70b_type-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}

llama2_7b_responsibility-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-7b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_7b_responsibility-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcd269e5a4b9108be77479e726a1cb4f7f6f5d420072113b7f8546073545986b
+size 67201802

llama2_7b_responsibility-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,20 @@
+{
+  "base_model_name": "meta-llama_Llama-2-7b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.2940330505371094,
+  "eval_runtime": 178.282,
+  "eval_samples_per_second": 8.666,
+  "eval_steps_per_second": 1.088,
+  "epoch": 4.0,
+  "current_steps": 4701,
+  "loss": 0.2961,
+  "learning_rate": 4.195804195804196e-06,
+  "train_runtime": 10950.2359,
+  "train_samples_per_second": 1.719,
+  "train_steps_per_second": 0.013,
+  "total_flos": 5.982058047995904e+17,
+  "train_loss": 0.4049693025699278
+}

llama2_7b_responsibility-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_7b_responsibility-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 800,
+  "dataset": "responsibility_train",
+  "eval_dataset": "responsibility_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_7b_responsibility-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}

llama2_7b_type-16R-128B-4E-100S/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "models/meta-llama_Llama-2-7b-chat-hf",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "o_proj",
+    "v_proj",
+    "k_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

llama2_7b_type-16R-128B-4E-100S/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:900dcca63d709a15775775098f1153a07990161a4bb6f9388393172d4f886f27
+size 67201802

llama2_7b_type-16R-128B-4E-100S/training_log.json

@@ -0,0 +1,20 @@
+{
+  "base_model_name": "meta-llama_Llama-2-7b-chat-hf",
+  "base_model_class": "LlamaForCausalLM",
+  "base_loaded_in_4bit": false,
+  "base_loaded_in_8bit": false,
+  "projections": "q, v, k, o",
+  "eval_loss": 0.7212664484977722,
+  "eval_runtime": 139.4052,
+  "eval_samples_per_second": 8.178,
+  "eval_steps_per_second": 1.026,
+  "epoch": 3.98,
+  "current_steps": 4586,
+  "loss": 0.7217,
+  "learning_rate": 8.571428571428571e-06,
+  "train_runtime": 11212.7879,
+  "train_samples_per_second": 1.653,
+  "train_steps_per_second": 0.013,
+  "total_flos": 7.497902781722788e+17,
+  "train_loss": 0.8539240459601084
+}

llama2_7b_type-16R-128B-4E-100S/training_parameters.json

@@ -0,0 +1,37 @@
+{
+  "lora_name": "llama2_7b_type-16R-128B-4E-100S",
+  "always_override": true,
+  "q_proj_en": true,
+  "v_proj_en": true,
+  "k_proj_en": true,
+  "o_proj_en": true,
+  "gate_proj_en": false,
+  "down_proj_en": false,
+  "up_proj_en": false,
+  "save_steps": 100.0,
+  "micro_batch_size": 4,
+  "batch_size": 128,
+  "epochs": 4.0,
+  "learning_rate": "3e-4",
+  "lr_scheduler_type": "linear",
+  "lora_rank": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "cutoff_len": 1024,
+  "dataset": "type_train",
+  "eval_dataset": "type_val",
+  "format": "alpaca-format",
+  "eval_steps": 100.0,
+  "raw_text_file": "None",
+  "overlap_len": 128,
+  "newline_favor_len": 128,
+  "higher_rank_limit": false,
+  "warmup_steps": 100.0,
+  "optimizer": "adamw_torch",
+  "hard_cut_string": "\\n\\n\\n",
+  "train_only_after": "",
+  "stop_at_loss": 0,
+  "add_eos_token": false,
+  "min_chars": 0.0,
+  "report_to": "tensorboard"
+}

llama2_7b_type-16R-128B-4E-100S/training_prompt.json

@@ -0,0 +1,5 @@
+{
+  "template_type": "dataset",
+  "template_1": "### Instruction:\n%instruction%\n\n### Response:\n%output%",
+  "template_2": "### Instruction:\n%instruction%\n\n### Input:\n%input%\n\n### Response:\n%output%"
+}