metadata

base_model: meta-llama/Llama-2-7b-hf
tags:
  - generated_from_trainer
model-index:
  - name: sparse_llama_7b_hf2_refined_web_50p_2024-03-27
    results: []

sparse_llama_7b_hf2_refined_web_50p_2024-03-27

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the None dataset. It achieves the following results on the evaluation set:

Loss: 2.0766

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 1
eval_batch_size: 4
seed: 0
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 8
total_train_batch_size: 32
total_eval_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
training_steps: 1100

Training results

Training Loss	Epoch	Step	Validation Loss
2.1821	0.01	25	2.2632
2.1824	0.02	50	2.2610
2.2916	0.02	75	2.2561
2.2562	0.03	100	2.2484
2.3387	0.04	125	2.2453
2.1762	0.05	150	2.2402
2.1439	0.06	175	2.2353
2.3081	0.06	200	2.2326
2.268	0.07	225	2.2300
2.2193	0.08	250	2.2303
2.1589	0.09	275	2.2296
2.1932	0.1	300	2.2276
2.2406	0.1	325	2.2271
2.2102	0.11	350	2.2289
2.1311	0.12	375	2.2272
2.2318	0.13	400	2.2269
2.2155	0.14	425	2.2273
2.1799	0.14	450	2.2267
2.252	0.15	475	2.2250
2.2588	0.16	500	2.2262
2.1677	0.17	525	2.2271
2.163	0.18	550	2.2264
2.2783	0.18	575	2.2251
2.1625	0.19	600	2.2253
2.1906	0.2	625	2.2251
2.2748	0.21	650	2.2251
2.171	0.22	675	2.2249
2.1929	0.22	700	2.2252
2.2203	0.23	725	2.2232
2.1143	0.24	750	2.2239
2.1969	0.25	775	2.2230
2.2492	0.26	800	2.2233
2.1988	0.26	825	2.2240
2.1546	0.27	850	2.2245
2.1605	0.28	875	2.2229
2.1417	0.29	900	2.2224
2.3172	0.3	925	2.2247
2.2799	0.3	950	2.2240
2.2258	0.31	975	2.2221
2.1175	0.32	1000	2.2216
2.2314	0.33	1025	2.2227
2.1956	0.34	1050	2.2211
2.1695	0.34	1075	2.2206
2.1658	0.35	1100	2.2207

Framework versions

Transformers 4.40.0.dev0
Pytorch 2.1.1+cu121
Datasets 2.15.0
Tokenizers 0.15.2