martimfasantos commited on
Commit
bc5dc81
1 Parent(s): 44b1dd3

End of training

Browse files
README.md CHANGED
@@ -2,10 +2,13 @@
2
  license: apache-2.0
3
  library_name: peft
4
  tags:
 
5
  - trl
6
  - dpo
7
  - generated_from_trainer
8
  base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
 
 
9
  model-index:
10
  - name: tinyllama-1.1b-chat-dpo-qlora
11
  results: []
@@ -16,15 +19,15 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # tinyllama-1.1b-chat-dpo-qlora
18
 
19
- This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T](https://huggingface.co/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.6085
22
- - Rewards/chosen: -1.0876
23
- - Rewards/rejected: -1.3914
24
  - Rewards/accuracies: 0.6580
25
- - Rewards/margins: 0.3038
26
- - Logps/rejected: -490.8211
27
- - Logps/chosen: -504.9807
28
  - Logits/rejected: -2.6096
29
  - Logits/chosen: -2.6425
30
 
 
2
  license: apache-2.0
3
  library_name: peft
4
  tags:
5
+ - alignment-handbook
6
  - trl
7
  - dpo
8
  - generated_from_trainer
9
  base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
10
+ datasets:
11
+ - HuggingFaceH4/ultrafeedback_binarized
12
  model-index:
13
  - name: tinyllama-1.1b-chat-dpo-qlora
14
  results: []
 
19
 
20
  # tinyllama-1.1b-chat-dpo-qlora
21
 
22
+ This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-chat-sft-qlora](https://huggingface.co/martimfasantos/tinyllama-1.1b-chat-sft-qlora) on the HuggingFaceH4/ultrafeedback_binarized dataset.
23
  It achieves the following results on the evaluation set:
24
+ - Loss: 0.6084
25
+ - Rewards/chosen: -1.0875
26
+ - Rewards/rejected: -1.3916
27
  - Rewards/accuracies: 0.6580
28
+ - Rewards/margins: 0.3041
29
+ - Logps/rejected: -490.8393
30
+ - Logps/chosen: -504.9714
31
  - Logits/rejected: -2.6096
32
  - Logits/chosen: -2.6425
33
 
all_results.json CHANGED
@@ -1,5 +1,18 @@
1
  {
2
  "epoch": 1.0,
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  "train_loss": 0.6288731582999011,
4
  "train_runtime": 37165.2285,
5
  "train_samples": 61134,
 
1
  {
2
  "epoch": 1.0,
3
+ "eval_logits/chosen": -2.642491102218628,
4
+ "eval_logits/rejected": -2.609631299972534,
5
+ "eval_logps/chosen": -504.97137451171875,
6
+ "eval_logps/rejected": -490.8393249511719,
7
+ "eval_loss": 0.6084445714950562,
8
+ "eval_rewards/accuracies": 0.6579999923706055,
9
+ "eval_rewards/chosen": -1.0874665975570679,
10
+ "eval_rewards/margins": 0.3040919303894043,
11
+ "eval_rewards/rejected": -1.3915585279464722,
12
+ "eval_runtime": 347.6592,
13
+ "eval_samples": 2000,
14
+ "eval_samples_per_second": 5.753,
15
+ "eval_steps_per_second": 0.719,
16
  "train_loss": 0.6288731582999011,
17
  "train_runtime": 37165.2285,
18
  "train_samples": 61134,
config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T",
3
+ "architectures": [
4
+ "LlamaForCausalLM"
5
+ ],
6
+ "attention_bias": false,
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 1,
9
+ "eos_token_id": 2,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 2048,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 5632,
14
+ "max_position_embeddings": 2048,
15
+ "model_type": "llama",
16
+ "num_attention_heads": 32,
17
+ "num_hidden_layers": 22,
18
+ "num_key_value_heads": 4,
19
+ "pretraining_tp": 1,
20
+ "quantization_config": {
21
+ "_load_in_4bit": true,
22
+ "_load_in_8bit": false,
23
+ "bnb_4bit_compute_dtype": "bfloat16",
24
+ "bnb_4bit_quant_storage": "uint8",
25
+ "bnb_4bit_quant_type": "nf4",
26
+ "bnb_4bit_use_double_quant": false,
27
+ "llm_int8_enable_fp32_cpu_offload": false,
28
+ "llm_int8_has_fp16_weight": false,
29
+ "llm_int8_skip_modules": null,
30
+ "llm_int8_threshold": 6.0,
31
+ "load_in_4bit": true,
32
+ "load_in_8bit": false,
33
+ "quant_method": "bitsandbytes"
34
+ },
35
+ "rms_norm_eps": 1e-05,
36
+ "rope_scaling": null,
37
+ "rope_theta": 10000.0,
38
+ "tie_word_embeddings": false,
39
+ "torch_dtype": "bfloat16",
40
+ "transformers_version": "4.39.3",
41
+ "use_cache": true,
42
+ "vocab_size": 32000
43
+ }
eval_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_logits/chosen": -2.642491102218628,
4
+ "eval_logits/rejected": -2.609631299972534,
5
+ "eval_logps/chosen": -504.97137451171875,
6
+ "eval_logps/rejected": -490.8393249511719,
7
+ "eval_loss": 0.6084445714950562,
8
+ "eval_rewards/accuracies": 0.6579999923706055,
9
+ "eval_rewards/chosen": -1.0874665975570679,
10
+ "eval_rewards/margins": 0.3040919303894043,
11
+ "eval_rewards/rejected": -1.3915585279464722,
12
+ "eval_runtime": 347.6592,
13
+ "eval_samples": 2000,
14
+ "eval_samples_per_second": 5.753,
15
+ "eval_steps_per_second": 0.719
16
+ }
runs/Apr24_01-31-15_poseidon/events.out.tfevents.1713959835.poseidon.732971.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26ee8116844081c9a4472a0ad53547d4f74c6585fc7112bb30015206797f68f0
3
+ size 828