ep1 packing v3

config.json

@@ -0,0 +1,22 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 3200,
+  "initializer_range": 0.02,
+  "intermediate_size": 8640,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 26,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.31.0.dev0",
+  "use_cache": true,
+  "vocab_size": 32000
+}

configs/packing-2.yml

@@ -0,0 +1,80 @@
+# base_model: meta-llama/Llama-2-7b-hf
+# base_model_config: meta-llama/Llama-2-7b-hf
+base_model: openlm-research/open_llama_3b_v2
+base_model_config: openlm-research/open_llama_3b_v2
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+tokenizer_use_fast: false  
+load_in_8bit: false                    
+load_in_4bit: false                                    
+strict: false  
+push_dataset_to_hub: openaccess-ai-collective
+hf_use_auth_token: true
+datasets:           
+  - path: Open-Orca/oo-gpt4-200k
+    type: alpaca_w_system.load_open_orca
+dataset_shard_num: 5
+dataset_shard_idx: 0
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.01
+adapter:
+lora_model_dir:
+sequence_len: 2048
+max_packed_sequence_len:
+sample_packing: true
+sample_packing_eff_est: 0.98
+lora_r:
+lora_alpha:
+lora_dropout:
+lora_target_modules:
+lora_target_linear:
+lora_fan_in_fan_out:
+wandb_project: packing-tests-3b
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+output_dir: ./open-orca-3b
+gradient_accumulation_steps: 1
+micro_batch_size: 4
+num_epochs: 6
+optimizer: adamw_torch
+adam_beta2: 0.95
+max_grad_norm: 1.0
+torchdistx_path:
+lr_scheduler: cosine
+lr_quadratic_warmup: true
+learning_rate: 0.0000045
+train_on_inputs: false
+group_by_length: false
+bf16: true
+fp16: false
+tf32: true
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention: 
+flash_attention:
+sdp_attention: true
+flash_optimum: 
+gptq_groupsize:
+gptq_model_v1:
+warmup_steps: 10
+eval_steps: 49
+save_steps: 
+debug:
+deepspeed:
+weight_decay: 0.1
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_offload_params: true
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
+

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.31.0.dev0"
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10150d89cbc8eee3fa245d3f0b3499e382232a27fb3d83263123f73292bda083
+size 6853066089

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": {"content": "<s>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}, "eos_token": {"content": "</s>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}, "unk_token": {"content": "<unk>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91b289e85fa20fd375d8b33dc12f77616f18abc6359804471d1fafcb425fecb8
+size 511574

tokenizer_config.json

@@ -0,0 +1 @@
+{"add_bos_token": true, "add_eos_token": false, "model_max_length": 2048, "pad_token": null, "sp_model_kwargs": {}, "tokenizer_class": "LlamaTokenizer", "clean_up_tokenization_spaces": false, "bos_token": {"__type": "AddedToken", "content": "<s>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}, "eos_token": {"__type": "AddedToken", "content": "</s>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}, "unk_token": {"__type": "AddedToken", "content": "<unk>", "lstrip": false, "normalized": true, "rstrip": false, "single_word": false}}