Add recipes/A5000_24GB_x8/coder-dolphin-en.yaml

Author: yuiseki

.vscode/settings.json

@@ -17,6 +17,7 @@
     "carblacac",
     "Cheung",
     "coefs",
+    "cognitivecomputations",
     "cosmopedia",
     "CSDS",
     "csebuetnlp",

recipes/A5000_24GB_x8/coder-dolphin-en.yaml

@@ -0,0 +1,33 @@
+target_task: tasks/nlp/translation.md
+base_model_id: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
+model_name: tinyllama-coder-dolphin-en-v0.1
+output_base_dir: /data/output
+dataset_id: cognitivecomputations/dolphin
+dataset_input_field_name: instruction
+dataset_context_field_name: input
+dataset_output_field_name: output
+dataset_train_split_seed: 42
+dataset_train_split_test_size: 0.2
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+train_claim_gpu_num: 3
+train_per_device_train_batch_size: 8
+train_gradient_accumulation_steps: 2
+train_num_train_epochs: 4
+train_max_steps: 1000
+train_fp16: True
+inference_max_new_tokens: 32
+evaluations:
+  -
+    prompt: "thank you"
+    expected_output: "ありがとう"
+  -
+    prompt: "Hello"
+    expected_output: "こんにちは"
+  -
+    prompt: "How are you?"
+    expected_output: "お元気ですか？"
+  -
+    prompt: "I am hungry"
+    expected_output: "お腹が空いています"