End of training

README.md

@@ -2,6 +2,7 @@
 base_model: princeton-nlp/Llama-3-Base-8B-SFT
 library_name: peft
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
@@ -18,15 +19,15 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [princeton-nlp/Llama-3-Base-8B-SFT](https://huggingface.co/princeton-nlp/Llama-3-Base-8B-SFT) on the None dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.5134
-- Rewards/chosen: -0.2023
-- Rewards/rejected: -1.1119
-- Rewards/accuracies: 0.7480
-- Rewards/margins: 0.9095
-- Logps/rejected: -287.7953
-- Logps/chosen: -294.5704
+- Rewards/chosen: -0.2002
+- Rewards/rejected: -1.1148
+- Rewards/accuracies: 0.7440
+- Rewards/margins: 0.9146
+- Logps/rejected: -287.8242
+- Logps/chosen: -294.5491
 - Logps/ref Response: -0.5364
-- Logits/rejected: -0.1602
-- Logits/chosen: -0.2100
+- Logits/rejected: -0.1599
+- Logits/chosen: -0.2095
 
 ## Model description
 

all_results.json

@@ -1,5 +1,19 @@
 {
     "epoch": 0.9997382884061764,
+    "eval_logits/chosen": -0.2095365673303604,
+    "eval_logits/rejected": -0.15986579656600952,
+    "eval_logps/chosen": -294.54913330078125,
+    "eval_logps/ref_response": -0.5363935232162476,
+    "eval_logps/rejected": -287.82421875,
+    "eval_loss": 0.5134016275405884,
+    "eval_rewards/accuracies": 0.7440000176429749,
+    "eval_rewards/chosen": -0.200197234749794,
+    "eval_rewards/margins": 0.9145647287368774,
+    "eval_rewards/rejected": -1.1147618293762207,
+    "eval_runtime": 348.7904,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 5.734,
+    "eval_steps_per_second": 0.358,
     "total_flos": 0.0,
     "train_loss": 0.5342667900454936,
     "train_runtime": 19113.1655,

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "princeton-nlp/Llama-3-Base-8B-SFT",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "vocab_size": 128256
+}

eval_results.json

@@ -0,0 +1,17 @@
+{
+    "epoch": 0.9997382884061764,
+    "eval_logits/chosen": -0.2095365673303604,
+    "eval_logits/rejected": -0.15986579656600952,
+    "eval_logps/chosen": -294.54913330078125,
+    "eval_logps/ref_response": -0.5363935232162476,
+    "eval_logps/rejected": -287.82421875,
+    "eval_loss": 0.5134016275405884,
+    "eval_rewards/accuracies": 0.7440000176429749,
+    "eval_rewards/chosen": -0.200197234749794,
+    "eval_rewards/margins": 0.9145647287368774,
+    "eval_rewards/rejected": -1.1147618293762207,
+    "eval_runtime": 348.7904,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 5.734,
+    "eval_steps_per_second": 0.358
+}