End of training

README.md

@@ -1,8 +1,10 @@
 ---
 base_model: Qwen/Qwen2-1.5B
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 library_name: peft
-license: apache-2.0
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
@@ -16,17 +18,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-dpo-qlora
 
-This model is a fine-tuned version of [Qwen/Qwen2-1.5B](https://huggingface.co/Qwen/Qwen2-1.5B) on an unknown dataset.
+This model is a fine-tuned version of [/opt/data/private/xgq/alignment-handbook/data/Qwen-1.5b-sft-qlora](https://huggingface.co//opt/data/private/xgq/alignment-handbook/data/Qwen-1.5b-sft-qlora) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.5488
-- Rewards/chosen: -1.1271
-- Rewards/rejected: -1.7888
+- Loss: 0.5487
+- Rewards/chosen: -1.1270
+- Rewards/rejected: -1.7889
 - Rewards/accuracies: 0.7380
-- Rewards/margins: 0.6617
-- Logps/rejected: -483.5208
-- Logps/chosen: -460.1232
-- Logits/rejected: -1.4135
-- Logits/chosen: -1.4626
+- Rewards/margins: 0.6620
+- Logps/rejected: -483.5314
+- Logps/chosen: -460.1111
+- Logits/rejected: -1.4133
+- Logits/chosen: -1.4624
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 0.9999018549416037,
+    "eval_logits/chosen": -1.4624338150024414,
+    "eval_logits/rejected": -1.4133074283599854,
+    "eval_logps/chosen": -460.1110534667969,
+    "eval_logps/rejected": -483.5314025878906,
+    "eval_loss": 0.5487077236175537,
+    "eval_rewards/accuracies": 0.7380239367485046,
+    "eval_rewards/chosen": -1.126965880393982,
+    "eval_rewards/margins": 0.6619576215744019,
+    "eval_rewards/rejected": -1.7889233827590942,
+    "eval_runtime": 296.9255,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 6.736,
+    "eval_steps_per_second": 1.125,
     "total_flos": 0.0,
     "train_loss": 0.5787653771390342,
     "train_runtime": 28047.1443,

config.json

@@ -0,0 +1,43 @@
+{
+  "_name_or_path": "Qwen/Qwen2-1.5B",
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "max_position_embeddings": 131072,
+  "max_window_layers": 28,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "quantization_config": {
+    "_load_in_4bit": true,
+    "_load_in_8bit": false,
+    "bnb_4bit_compute_dtype": "bfloat16",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_use_double_quant": false,
+    "llm_int8_enable_fp32_cpu_offload": false,
+    "llm_int8_has_fp16_weight": false,
+    "llm_int8_skip_modules": null,
+    "llm_int8_threshold": 6.0,
+    "load_in_4bit": true,
+    "load_in_8bit": false,
+    "quant_method": "bitsandbytes"
+  },
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.44.2",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 0.9999018549416037,
+    "eval_logits/chosen": -1.4624338150024414,
+    "eval_logits/rejected": -1.4133074283599854,
+    "eval_logps/chosen": -460.1110534667969,
+    "eval_logps/rejected": -483.5314025878906,
+    "eval_loss": 0.5487077236175537,
+    "eval_rewards/accuracies": 0.7380239367485046,
+    "eval_rewards/chosen": -1.126965880393982,
+    "eval_rewards/margins": 0.6619576215744019,
+    "eval_rewards/rejected": -1.7889233827590942,
+    "eval_runtime": 296.9255,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 6.736,
+    "eval_steps_per_second": 1.125
+}

runs/Sep26_15-25-38_interactive79915/events.out.tfevents.1727364526.interactive79915.142135.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0bd48f7da00e53873826225d79d18630d6b45b45b938d23540e625893f39daf
+size 828