Skip to content

Commit 8145b36

Browse files
mresomreso
committed
Added cell outputs
1 parent ddfce80 commit 8145b36

File tree

1 file changed

+166
-14
lines changed

1 file changed

+166
-14
lines changed

‎recipes/finetuning/quickstart_peft_finetuning.ipynb

Lines changed: 166 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,7 @@
3030
},
3131
{
3232
"cell_type": "code",
33-
"execution_count": null,
33+
"execution_count": 1,
3434
"metadata": {},
3535
"outputs": [],
3636
"source": [
@@ -52,9 +52,31 @@
5252
},
5353
{
5454
"cell_type": "code",
55-
"execution_count": null,
55+
"execution_count": 2,
5656
"metadata": {},
57-
"outputs": [],
57+
"outputs": [
58+
{
59+
"data": {
60+
"application/vnd.jupyter.widget-view+json": {
61+
"model_id": "c7963d43806d432aaa3d00e2055e355c",
62+
"version_major": 2,
63+
"version_minor": 0
64+
},
65+
"text/plain": [
66+
"Loading checkpoint shards: 0%| | 0/4 [00:00<?, ?it/s]"
67+
]
68+
},
69+
"metadata": {},
70+
"output_type": "display_data"
71+
},
72+
{
73+
"name": "stderr",
74+
"output_type": "stream",
75+
"text": [
76+
"Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n"
77+
]
78+
}
79+
],
5880
"source": [
5981
"import torch\n",
6082
"from transformers import LlamaForCausalLM, AutoTokenizer\n",
@@ -103,9 +125,52 @@
103125
},
104126
{
105127
"cell_type": "code",
106-
"execution_count": null,
128+
"execution_count": 3,
107129
"metadata": {},
108-
"outputs": [],
130+
"outputs": [
131+
{
132+
"name": "stderr",
133+
"output_type": "stream",
134+
"text": [
135+
"Setting `pad_token_id` to `eos_token_id`:128001 for open-end generation.\n"
136+
]
137+
},
138+
{
139+
"name": "stdout",
140+
"output_type": "stream",
141+
"text": [
142+
"\n",
143+
"Summarize this dialog:\n",
144+
"A: Hi Tom, are you busy tomorrow’s afternoon?\n",
145+
"B: I’m pretty sure I am. What’s up?\n",
146+
"A: Can you go with me to the animal shelter?.\n",
147+
"B: What do you want to do?\n",
148+
"A: I want to get a puppy for my son.\n",
149+
"B: That will make him so happy.\n",
150+
"A: Yeah, we’ve discussed it many times. I think he’s ready now.\n",
151+
"B: That’s good. Raising a dog is a tough issue. Like having a baby ;-) \n",
152+
"A: I'll get him one of those little dogs.\n",
153+
"B: One that won't grow up too big;-)\n",
154+
"A: And eat too much;-))\n",
155+
"B: Do you know which one he would like?\n",
156+
"A: Oh, yes, I took him there last Monday. He showed me one that he really liked.\n",
157+
"B: I bet you had to drag him away.\n",
158+
"A: He wanted to take it home right away ;-).\n",
159+
"B: I wonder what he'll name it.\n",
160+
"A: He said he’d name it after his dead hamster – Lemmy - he's a great Motorhead fan :-)))\n",
161+
"---\n",
162+
"Summary:\n",
163+
"A: Hi Tom, are you busy tomorrow’s afternoon?\n",
164+
"B: I’m pretty sure I am. What’s up?\n",
165+
"A: Can you go with me to the animal shelter?.\n",
166+
"B: What do you want to do?\n",
167+
"A: I want to get a puppy for my son.\n",
168+
"B: That will make him so happy.\n",
169+
"A: Yeah, we’ve discussed it many times. I think he’s ready now.\n",
170+
"B: That’s good. Raising a dog is a tough issue\n"
171+
]
172+
}
173+
],
109174
"source": [
110175
"eval_prompt = \"\"\"\n",
111176
"Summarize this dialog:\n",
@@ -151,9 +216,21 @@
151216
},
152217
{
153218
"cell_type": "code",
154-
"execution_count": null,
219+
"execution_count": 4,
155220
"metadata": {},
156-
"outputs": [],
221+
"outputs": [
222+
{
223+
"name": "stderr",
224+
"output_type": "stream",
225+
"text": [
226+
"/home/ubuntu/miniconda3/envs/llama/lib/python3.11/site-packages/datasets/load.py:1486: FutureWarning: The repository for samsum contains custom code which must be executed to correctly load the dataset. You can inspect the repository content at https://hf.co/datasets/samsum\n",
227+
"You can avoid this message in future by passing the argument `trust_remote_code=True`.\n",
228+
"Passing `trust_remote_code=True` will be mandatory to load this dataset from the next major release of `datasets`.\n",
229+
" warnings.warn(\n",
230+
"Preprocessing dataset: 100%|██████████| 14732/14732 [00:02<00:00, 6124.69it/s]\n"
231+
]
232+
}
233+
],
157234
"source": [
158235
"from llama_recipes.configs.datasets import samsum_dataset\n",
159236
"from llama_recipes.data.concatenator import ConcatDataset\n",
@@ -188,7 +265,7 @@
188265
},
189266
{
190267
"cell_type": "code",
191-
"execution_count": null,
268+
"execution_count": 5,
192269
"metadata": {},
193270
"outputs": [],
194271
"source": [
@@ -219,9 +296,39 @@
219296
},
220297
{
221298
"cell_type": "code",
222-
"execution_count": null,
299+
"execution_count": 6,
223300
"metadata": {},
224-
"outputs": [],
301+
"outputs": [
302+
{
303+
"name": "stderr",
304+
"output_type": "stream",
305+
"text": [
306+
"/home/ubuntu/miniconda3/envs/llama/lib/python3.11/site-packages/torch/cuda/memory.py:330: FutureWarning: torch.cuda.reset_max_memory_allocated now calls torch.cuda.reset_peak_memory_stats, which resets /all/ peak memory stats.\n",
307+
" warnings.warn(\n",
308+
"Training Epoch: 1: 0%|\u001b[34m \u001b[0m| 0/319 [00:00<?, ?it/s]huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...\n",
309+
"To disable this warning, you can either:\n",
310+
"\t- Avoid using `tokenizers` before the fork if possible\n",
311+
"\t- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)\n",
312+
"/home/ubuntu/miniconda3/envs/llama/lib/python3.11/site-packages/torch/utils/checkpoint.py:464: UserWarning: torch.utils.checkpoint: the use_reentrant parameter should be passed explicitly. In version 2.4 we will raise an exception if use_reentrant is not passed. use_reentrant=False is recommended, but if you need to preserve the current default behavior, you can pass use_reentrant=True. Refer to docs for more details on the differences between the two variants.\n",
313+
" warnings.warn(\n",
314+
"/home/ubuntu/miniconda3/envs/llama/lib/python3.11/site-packages/bitsandbytes/autograd/_functions.py:316: UserWarning: MatMul8bitLt: inputs will be cast from torch.float32 to float16 during quantization\n",
315+
" warnings.warn(f\"MatMul8bitLt: inputs will be cast from {A.dtype} to float16 during quantization\")\n",
316+
"Training Epoch: 1/1, step 1278/1279 completed (loss: 0.27870458364486694): : 320it [2:07:09, 23.84s/it] 3.94s/it] \n"
317+
]
318+
},
319+
{
320+
"name": "stdout",
321+
"output_type": "stream",
322+
"text": [
323+
"Max CUDA memory allocated was 15 GB\n",
324+
"Max CUDA memory reserved was 16 GB\n",
325+
"Peak active CUDA memory was 15 GB\n",
326+
"CUDA Malloc retries : 0\n",
327+
"CPU Total Peak Memory consumed during the train (max): 2 GB\n",
328+
"Epoch 1: train_perplexity=1.3403, train_epoch_loss=0.2929, epoch time 7630.169942979002s\n"
329+
]
330+
}
331+
],
225332
"source": [
226333
"import torch.optim as optim\n",
227334
"from llama_recipes.utils.train_utils import train\n",
@@ -264,9 +371,18 @@
264371
},
265372
{
266373
"cell_type": "code",
267-
"execution_count": null,
374+
"execution_count": 7,
268375
"metadata": {},
269-
"outputs": [],
376+
"outputs": [
377+
{
378+
"name": "stderr",
379+
"output_type": "stream",
380+
"text": [
381+
"/home/ubuntu/miniconda3/envs/llama/lib/python3.11/site-packages/huggingface_hub/file_download.py:1132: FutureWarning: `resume_download` is deprecated and will be removed in version 1.0.0. Downloads always resume when possible. If you want to force a new download, use `force_download=True`.\n",
382+
" warnings.warn(\n"
383+
]
384+
}
385+
],
270386
"source": [
271387
"model.save_pretrained(train_config.output_dir)"
272388
]
@@ -282,9 +398,45 @@
282398
},
283399
{
284400
"cell_type": "code",
285-
"execution_count": null,
401+
"execution_count": 8,
286402
"metadata": {},
287-
"outputs": [],
403+
"outputs": [
404+
{
405+
"name": "stderr",
406+
"output_type": "stream",
407+
"text": [
408+
"Setting `pad_token_id` to `eos_token_id`:128001 for open-end generation.\n"
409+
]
410+
},
411+
{
412+
"name": "stdout",
413+
"output_type": "stream",
414+
"text": [
415+
"\n",
416+
"Summarize this dialog:\n",
417+
"A: Hi Tom, are you busy tomorrow’s afternoon?\n",
418+
"B: I’m pretty sure I am. What’s up?\n",
419+
"A: Can you go with me to the animal shelter?.\n",
420+
"B: What do you want to do?\n",
421+
"A: I want to get a puppy for my son.\n",
422+
"B: That will make him so happy.\n",
423+
"A: Yeah, we’ve discussed it many times. I think he’s ready now.\n",
424+
"B: That’s good. Raising a dog is a tough issue. Like having a baby ;-) \n",
425+
"A: I'll get him one of those little dogs.\n",
426+
"B: One that won't grow up too big;-)\n",
427+
"A: And eat too much;-))\n",
428+
"B: Do you know which one he would like?\n",
429+
"A: Oh, yes, I took him there last Monday. He showed me one that he really liked.\n",
430+
"B: I bet you had to drag him away.\n",
431+
"A: He wanted to take it home right away ;-).\n",
432+
"B: I wonder what he'll name it.\n",
433+
"A: He said he’d name it after his dead hamster – Lemmy - he's a great Motorhead fan :-)))\n",
434+
"---\n",
435+
"Summary:\n",
436+
"A wants to get a puppy for her son. She will take him to the animal shelter tomorrow. B is not sure if he can go with her, but he's willing to.\n"
437+
]
438+
}
439+
],
288440
"source": [
289441
"model.eval()\n",
290442
"with torch.no_grad():\n",

0 commit comments

Comments
 (0)