Upload 9 files

robin-7b/README.md

@@ -1,20 +1,20 @@
 ---
-license: other
+license: apache-2.0
 tags:
 - generated_from_trainer
 datasets:
 - customized
 model-index:
-- name: xl_031_lora
+- name: h10
   results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# xl_031_lora
+# h10
 
-This model is a fine-tuned version of [aleksickx/llama-7b-hf](https://huggingface.co/aleksickx/llama-7b-hf) on the customized dataset.
+This model is a fine-tuned version of [pinkmanlove/llama-7b-hf](https://huggingface.co/pinkmanlove/llama-7b-hf) on the customized dataset.
 
 ## Model description
 
@@ -33,19 +33,18 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0008
+- learning_rate: 0.0001
 - train_batch_size: 4
 - eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
-- num_devices: 8
-- gradient_accumulation_steps: 4
-- total_train_batch_size: 128
-- total_eval_batch_size: 64
+- num_devices: 4
+- total_train_batch_size: 16
+- total_eval_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.03
-- num_epochs: 10.0
+- num_epochs: 5.0
 
 ### Training results
 

robin-7b/adapter_config.json

@@ -1,5 +1,5 @@
 {
-  "base_model_name_or_path": "aleksickx/llama-7b-hf",
+  "base_model_name_or_path": "pinkmanlove/llama-7b-hf",
   "bias": "none",
   "enable_lora": null,
   "fan_in_fan_out": false,
@@ -10,10 +10,12 @@
   "merge_weights": false,
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 32,
+  "r": 128,
   "target_modules": [
     "q_proj",
-    "v_proj"
+    "k_proj",
+    "v_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

robin-7b/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:320829da8b0ace9f5ada5155c6cb119b8e4c03d9c507c0103ee4765fd8c47d83
-size 33574781
+oid sha256:250ed2611b7e1071a390509be29f05e25d5ecb2c703955b97b2a640ddd6ce337
+size 268476157

robin-7b/all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.99,
-    "train_loss": 0.8784972819876163,
-    "train_runtime": 13393.5014,
-    "train_samples": 72257,
-    "train_samples_per_second": 53.949,
-    "train_steps_per_second": 0.421
+    "epoch": 5.0,
+    "train_loss": 0.9734652058462079,
+    "train_runtime": 41271.4588,
+    "train_samples": 142397,
+    "train_samples_per_second": 17.251,
+    "train_steps_per_second": 1.078
 }

robin-7b/tokenizer_config.json

@@ -3,7 +3,7 @@
   "clean_up_tokenization_spaces": false,
   "eos_token": "",
   "model_max_length": 1000000000000000019884624838656,
-  "special_tokens_map_file": "/home/xiangliu/.cache/huggingface/hub/models--aleksickx--llama-7b-hf/snapshots/d7d132438caf5e95800f35dfc46cf82c2be9b365/special_tokens_map.json",
+  "special_tokens_map_file": "/root/data/.cache/hub/models--pinkmanlove--llama-7b-hf/snapshots/b3cde76468bad3c085ead29707ee7481121a4ca0/special_tokens_map.json",
   "tokenizer_class": "LlamaTokenizer",
   "unk_token": ""
 }

robin-7b/train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.99,
-    "train_loss": 0.8784972819876163,
-    "train_runtime": 13393.5014,
-    "train_samples": 72257,
-    "train_samples_per_second": 53.949,
-    "train_steps_per_second": 0.421
+    "epoch": 5.0,
+    "train_loss": 0.9734652058462079,
+    "train_runtime": 41271.4588,
+    "train_samples": 142397,
+    "train_samples_per_second": 17.251,
+    "train_steps_per_second": 1.078
 }