Instructions to use keepsloading/icml_repro_scratch with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use keepsloading/icml_repro_scratch with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="keepsloading/icml_repro_scratch")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("keepsloading/icml_repro_scratch")
model = AutoModelForCausalLM.from_pretrained("keepsloading/icml_repro_scratch", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use keepsloading/icml_repro_scratch with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "keepsloading/icml_repro_scratch"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "keepsloading/icml_repro_scratch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/keepsloading/icml_repro_scratch

SGLang

How to use keepsloading/icml_repro_scratch with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "keepsloading/icml_repro_scratch" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "keepsloading/icml_repro_scratch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "keepsloading/icml_repro_scratch" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "keepsloading/icml_repro_scratch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use keepsloading/icml_repro_scratch with Docker Model Runner:
```
docker model run hf.co/keepsloading/icml_repro_scratch
```

icml_repro_scratch / outputs /eval /logs /token_t045 /babilong.log

keepsloading

Upload folder using huggingface_hub

6a26953 verified 30 days ago

Raw

History Blame Contribute Delete

31 kB

	2026-07-18:15:17:55 WARNING [config.evaluate_config:281] --limit SHOULD ONLY BE USED FOR TESTING. REAL METRICS SHOULD NOT BE COMPUTED USING LIMIT.
	2026-07-18:15:17:58 INFO [_cli.run:376] Selected Tasks: ['babilong_longctx']
	2026-07-18:15:18:00 INFO [evaluator:211] Setting random seed to 0 \| Setting numpy seed to 1234 \| Setting torch manual seed to 1234 \| Setting fewshot manual seed to 1234
	2026-07-18:15:18:00 INFO [evaluator:236] Initializing hf model, with arguments: {'pretrained': '/workspace/outputs/l2a_style/stage2/checkpoint-25', 'trust_remote_code': True, 'dtype': 'bfloat16', 'max_length': 16384, 'attn_implementation': 'sdpa'}
	2026-07-18:15:18:03 INFO [models.huggingface:161] Using device 'cuda:0'
	2026-07-18:15:18:03 INFO [models.huggingface:548] Model type cannot be determined. Using default model type 'causal'
	2026-07-18:15:18:03 INFO [models.huggingface:423] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'}
	2026-07-18:15:18:05 WARNING [api.task:856] babilong_qa5: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager.
	For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme.
	2026-07-18:15:18:05 WARNING [api.task:856] babilong_qa4: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager.
	For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme.
	2026-07-18:15:18:05 WARNING [api.task:856] babilong_qa3: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager.
	For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme.
	2026-07-18:15:18:06 WARNING [api.task:856] babilong_qa2: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager.
	For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme.
	2026-07-18:15:18:06 WARNING [api.task:856] babilong_qa1: Custom kwargs can be passed to `--metadata` in console (as json string) or to the TaskManager.
	For example --metadata='{"max_seq_lengths":[4096, 8192]}'. For details see task Readme.
	2026-07-18:15:18:06 INFO [tasks:700] Selected tasks:
	2026-07-18:15:18:06 INFO [tasks:703] Group: babilong_longctx
	2026-07-18:15:18:06 INFO [tasks:726] ConfigurableGroup(group=babilong_longctx,group_alias=None): {'babilong_qa1': ConfigurableTask(task_name=babilong_qa1,output_type=generate_until,num_fewshot=2,num_samples=1000), 'babilong_qa2': ConfigurableTask(task_name=babilong_qa2,output_type=generate_until,num_fewshot=2,num_samples=999), 'babilong_qa3': ConfigurableTask(task_name=babilong_qa3,output_type=generate_until,num_fewshot=2,num_samples=999), 'babilong_qa4': ConfigurableTask(task_name=babilong_qa4,output_type=generate_until,num_fewshot=2,num_samples=999), 'babilong_qa5': ConfigurableTask(task_name=babilong_qa5,output_type=generate_until,num_fewshot=2,num_samples=999)}
	2026-07-18:15:18:06 INFO [evaluator:314] babilong_qa1: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 16, 'until': []}
	2026-07-18:15:18:06 WARNING [evaluator:333] Overwriting default num_fewshot of babilong_qa1 from 2 to 2
	2026-07-18:15:18:06 INFO [evaluator:314] babilong_qa2: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 16, 'until': []}
	2026-07-18:15:18:06 WARNING [evaluator:333] Overwriting default num_fewshot of babilong_qa2 from 2 to 2
	2026-07-18:15:18:06 INFO [evaluator:314] babilong_qa3: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 16, 'until': []}
	2026-07-18:15:18:06 WARNING [evaluator:333] Overwriting default num_fewshot of babilong_qa3 from 2 to 2
	2026-07-18:15:18:06 INFO [evaluator:314] babilong_qa4: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 16, 'until': []}
	2026-07-18:15:18:06 WARNING [evaluator:333] Overwriting default num_fewshot of babilong_qa4 from 2 to 2
	2026-07-18:15:18:06 INFO [evaluator:314] babilong_qa5: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 16, 'until': []}
	2026-07-18:15:18:06 WARNING [evaluator:333] Overwriting default num_fewshot of babilong_qa5 from 2 to 2
	2026-07-18:15:18:06 INFO [api.task:311] Building contexts for babilong_qa1 on rank 0...

	0%\| \| 0/50 [00:00<?, ?it/s][A 100%\|██████████\| 50/50 [00:00<00:00, 375.50it/s]
	2026-07-18:15:18:06 INFO [api.task:311] Building contexts for babilong_qa2 on rank 0...

	0%\| \| 0/50 [00:00<?, ?it/s][A 100%\|██████████\| 50/50 [00:00<00:00, 370.78it/s]
	2026-07-18:15:18:06 INFO [api.task:311] Building contexts for babilong_qa3 on rank 0...

	0%\| \| 0/50 [00:00<?, ?it/s][A 100%\|██████████\| 50/50 [00:00<00:00, 369.97it/s]
	2026-07-18:15:18:07 INFO [api.task:311] Building contexts for babilong_qa4 on rank 0...

	0%\| \| 0/50 [00:00<?, ?it/s][A 100%\|██████████\| 50/50 [00:00<00:00, 374.34it/s]
	2026-07-18:15:18:07 INFO [api.task:311] Building contexts for babilong_qa5 on rank 0...

	0%\| \| 0/50 [00:00<?, ?it/s][A 100%\|██████████\| 50/50 [00:00<00:00, 372.53it/s]
	2026-07-18:15:18:07 INFO [evaluator:584] Running generate_until requests

	Running generate_until requests: 0%\| \| 0/250 [00:00<?, ?it/s][A
	Running generate_until requests: 0%\| \| 1/250 [00:06<25:38, 6.18s/it][A
	Running generate_until requests: 1%\| \| 2/250 [00:07<13:51, 3.35s/it][A
	Running generate_until requests: 1%\| \| 3/250 [00:08<10:05, 2.45s/it][A
	Running generate_until requests: 2%\|▏ \| 4/250 [00:10<08:17, 2.02s/it][A
	Running generate_until requests: 2%\|▏ \| 5/250 [00:11<07:16, 1.78s/it][A
	Running generate_until requests: 2%\|▏ \| 6/250 [00:13<06:39, 1.64s/it][A
	Running generate_until requests: 3%\|▎ \| 7/250 [00:14<06:15, 1.55s/it][A
	Running generate_until requests: 3%\|▎ \| 8/250 [00:15<05:59, 1.49s/it][A
	Running generate_until requests: 4%\|▎ \| 9/250 [00:17<05:47, 1.44s/it][A
	Running generate_until requests: 4%\|▍ \| 10/250 [00:18<05:39, 1.42s/it][A
	Running generate_until requests: 4%\|▍ \| 11/250 [00:19<05:33, 1.40s/it][A
	Running generate_until requests: 5%\|▍ \| 12/250 [00:21<05:28, 1.38s/it][A
	Running generate_until requests: 5%\|▌ \| 13/250 [00:22<05:24, 1.37s/it][A
	Running generate_until requests: 6%\|▌ \| 14/250 [00:23<05:21, 1.36s/it][A
	Running generate_until requests: 6%\|▌ \| 15/250 [00:25<05:19, 1.36s/it][A
	Running generate_until requests: 6%\|▋ \| 16/250 [00:26<05:17, 1.36s/it][A
	Running generate_until requests: 7%\|▋ \| 17/250 [00:27<05:15, 1.35s/it][A
	Running generate_until requests: 7%\|▋ \| 18/250 [00:29<05:12, 1.35s/it][A
	Running generate_until requests: 8%\|▊ \| 19/250 [00:30<05:11, 1.35s/it][A
	Running generate_until requests: 8%\|▊ \| 20/250 [00:31<05:09, 1.35s/it][A
	Running generate_until requests: 8%\|▊ \| 21/250 [00:33<05:08, 1.35s/it][A
	Running generate_until requests: 9%\|▉ \| 22/250 [00:34<05:06, 1.34s/it][A
	Running generate_until requests: 9%\|▉ \| 23/250 [00:35<05:05, 1.35s/it][A
	Running generate_until requests: 10%\|▉ \| 24/250 [00:37<05:04, 1.35s/it][A
	Running generate_until requests: 10%\|█ \| 25/250 [00:38<05:02, 1.35s/it][A
	Running generate_until requests: 10%\|█ \| 26/250 [00:39<05:01, 1.35s/it][A
	Running generate_until requests: 11%\|█ \| 27/250 [00:41<05:00, 1.35s/it][A
	Running generate_until requests: 11%\|█ \| 28/250 [00:42<04:58, 1.35s/it][A
	Running generate_until requests: 12%\|█▏ \| 29/250 [00:43<04:57, 1.34s/it][A
	Running generate_until requests: 12%\|█▏ \| 30/250 [00:45<04:55, 1.34s/it][A
	Running generate_until requests: 12%\|█▏ \| 31/250 [00:46<04:54, 1.34s/it][A
	Running generate_until requests: 13%\|█▎ \| 32/250 [00:48<04:53, 1.34s/it][A
	Running generate_until requests: 13%\|█▎ \| 33/250 [00:49<04:51, 1.35s/it][A
	Running generate_until requests: 14%\|█▎ \| 34/250 [00:50<04:50, 1.34s/it][A
	Running generate_until requests: 14%\|█▍ \| 35/250 [00:52<04:49, 1.34s/it][A
	Running generate_until requests: 14%\|█▍ \| 36/250 [00:53<04:47, 1.34s/it][A
	Running generate_until requests: 15%\|█▍ \| 37/250 [00:54<04:46, 1.34s/it][A
	Running generate_until requests: 15%\|█▌ \| 38/250 [00:56<04:45, 1.34s/it][A
	Running generate_until requests: 16%\|█▌ \| 39/250 [00:57<04:44, 1.35s/it][A
	Running generate_until requests: 16%\|█▌ \| 40/250 [00:58<04:42, 1.35s/it][A
	Running generate_until requests: 16%\|█▋ \| 41/250 [01:00<04:41, 1.34s/it][A
	Running generate_until requests: 17%\|█▋ \| 42/250 [01:01<04:39, 1.34s/it][A
	Running generate_until requests: 17%\|█▋ \| 43/250 [01:02<04:38, 1.34s/it][A
	Running generate_until requests: 18%\|█▊ \| 44/250 [01:04<04:36, 1.34s/it][A
	Running generate_until requests: 18%\|█▊ \| 45/250 [01:05<04:35, 1.34s/it][A
	Running generate_until requests: 18%\|█▊ \| 46/250 [01:06<04:35, 1.35s/it][A
	Running generate_until requests: 19%\|█▉ \| 47/250 [01:08<04:34, 1.35s/it][A
	Running generate_until requests: 19%\|█▉ \| 48/250 [01:09<04:33, 1.35s/it][A
	Running generate_until requests: 20%\|█▉ \| 49/250 [01:10<04:31, 1.35s/it][A
	Running generate_until requests: 20%\|██ \| 50/250 [01:12<04:29, 1.35s/it][A
	Running generate_until requests: 20%\|██ \| 51/250 [01:13<04:28, 1.35s/it][A
	Running generate_until requests: 21%\|██ \| 52/250 [01:14<04:26, 1.35s/it][A
	Running generate_until requests: 21%\|██ \| 53/250 [01:16<04:25, 1.35s/it][A
	Running generate_until requests: 22%\|██▏ \| 54/250 [01:17<04:23, 1.35s/it][A
	Running generate_until requests: 22%\|██▏ \| 55/250 [01:18<04:22, 1.35s/it][A
	Running generate_until requests: 22%\|██▏ \| 56/250 [01:20<04:20, 1.34s/it][A
	Running generate_until requests: 23%\|██▎ \| 57/250 [01:21<04:19, 1.34s/it][A
	Running generate_until requests: 23%\|██▎ \| 58/250 [01:23<04:17, 1.34s/it][A
	Running generate_until requests: 24%\|██▎ \| 59/250 [01:24<04:15, 1.34s/it][A
	Running generate_until requests: 24%\|██▍ \| 60/250 [01:25<04:13, 1.34s/it][A
	Running generate_until requests: 24%\|██▍ \| 61/250 [01:27<04:12, 1.34s/it][A
	Running generate_until requests: 25%\|██▍ \| 62/250 [01:28<04:11, 1.34s/it][A
	Running generate_until requests: 25%\|██▌ \| 63/250 [01:29<04:10, 1.34s/it][A
	Running generate_until requests: 26%\|██▌ \| 64/250 [01:31<04:09, 1.34s/it][A
	Running generate_until requests: 26%\|██▌ \| 65/250 [01:32<04:07, 1.34s/it][A
	Running generate_until requests: 26%\|██▋ \| 66/250 [01:33<04:06, 1.34s/it][A
	Running generate_until requests: 27%\|██▋ \| 67/250 [01:35<04:04, 1.34s/it][A
	Running generate_until requests: 27%\|██▋ \| 68/250 [01:36<04:03, 1.34s/it][A
	Running generate_until requests: 28%\|██▊ \| 69/250 [01:37<04:02, 1.34s/it][A
	Running generate_until requests: 28%\|██▊ \| 70/250 [01:39<04:00, 1.34s/it][A
	Running generate_until requests: 28%\|██▊ \| 71/250 [01:40<03:59, 1.34s/it][A
	Running generate_until requests: 29%\|██▉ \| 72/250 [01:41<03:58, 1.34s/it][A
	Running generate_until requests: 29%\|██▉ \| 73/250 [01:43<03:57, 1.34s/it][A
	Running generate_until requests: 30%\|██▉ \| 74/250 [01:44<03:55, 1.34s/it][A
	Running generate_until requests: 30%\|███ \| 75/250 [01:45<03:54, 1.34s/it][A
	Running generate_until requests: 30%\|███ \| 76/250 [01:47<03:53, 1.34s/it][A
	Running generate_until requests: 31%\|███ \| 77/250 [01:48<03:52, 1.34s/it][A
	Running generate_until requests: 31%\|███ \| 78/250 [01:49<03:50, 1.34s/it][A
	Running generate_until requests: 32%\|███▏ \| 79/250 [01:51<03:49, 1.34s/it][A
	Running generate_until requests: 32%\|███▏ \| 80/250 [01:52<03:47, 1.34s/it][A
	Running generate_until requests: 32%\|███▏ \| 81/250 [01:53<03:46, 1.34s/it][A
	Running generate_until requests: 33%\|███▎ \| 82/250 [01:55<03:44, 1.34s/it][A
	Running generate_until requests: 33%\|███▎ \| 83/250 [01:56<03:43, 1.34s/it][A
	Running generate_until requests: 34%\|███▎ \| 84/250 [01:57<03:42, 1.34s/it][A
	Running generate_until requests: 34%\|███▍ \| 85/250 [01:59<03:40, 1.34s/it][A
	Running generate_until requests: 34%\|███▍ \| 86/250 [02:00<03:39, 1.34s/it][A
	Running generate_until requests: 35%\|███▍ \| 87/250 [02:01<03:38, 1.34s/it][A
	Running generate_until requests: 35%\|███▌ \| 88/250 [02:03<03:36, 1.34s/it][A
	Running generate_until requests: 36%\|███▌ \| 89/250 [02:04<03:35, 1.34s/it][A
	Running generate_until requests: 36%\|███▌ \| 90/250 [02:05<03:34, 1.34s/it][A
	Running generate_until requests: 36%\|███▋ \| 91/250 [02:07<03:33, 1.34s/it][A
	Running generate_until requests: 37%\|███▋ \| 92/250 [02:08<03:31, 1.34s/it][A
	Running generate_until requests: 37%\|███▋ \| 93/250 [02:09<03:30, 1.34s/it][A
	Running generate_until requests: 38%\|███▊ \| 94/250 [02:11<03:28, 1.34s/it][A
	Running generate_until requests: 38%\|███▊ \| 95/250 [02:12<03:27, 1.34s/it][A
	Running generate_until requests: 38%\|███▊ \| 96/250 [02:13<03:25, 1.34s/it][A
	Running generate_until requests: 39%\|███▉ \| 97/250 [02:15<03:24, 1.33s/it][A
	Running generate_until requests: 39%\|███▉ \| 98/250 [02:16<03:22, 1.33s/it][A
	Running generate_until requests: 40%\|███▉ \| 99/250 [02:17<03:21, 1.33s/it][A
	Running generate_until requests: 40%\|████ \| 100/250 [02:19<03:19, 1.33s/it][A
	Running generate_until requests: 40%\|████ \| 101/250 [02:20<03:18, 1.33s/it][A
	Running generate_until requests: 41%\|████ \| 102/250 [02:21<03:16, 1.33s/it][A
	Running generate_until requests: 41%\|████ \| 103/250 [02:23<03:15, 1.33s/it][A
	Running generate_until requests: 42%\|████▏ \| 104/250 [02:24<03:14, 1.33s/it][A
	Running generate_until requests: 42%\|████▏ \| 105/250 [02:25<03:12, 1.33s/it][A
	Running generate_until requests: 42%\|████▏ \| 106/250 [02:27<03:11, 1.33s/it][A
	Running generate_until requests: 43%\|████▎ \| 107/250 [02:28<03:10, 1.33s/it][A
	Running generate_until requests: 43%\|████▎ \| 108/250 [02:29<03:08, 1.33s/it][A
	Running generate_until requests: 44%\|████▎ \| 109/250 [02:31<03:07, 1.33s/it][A
	Running generate_until requests: 44%\|████▍ \| 110/250 [02:32<03:06, 1.33s/it][A
	Running generate_until requests: 44%\|████▍ \| 111/250 [02:33<03:04, 1.33s/it][A
	Running generate_until requests: 45%\|████▍ \| 112/250 [02:35<03:03, 1.33s/it][A
	Running generate_until requests: 45%\|████▌ \| 113/250 [02:36<03:02, 1.33s/it][A
	Running generate_until requests: 46%\|████▌ \| 114/250 [02:37<03:00, 1.33s/it][A
	Running generate_until requests: 46%\|████▌ \| 115/250 [02:39<02:59, 1.33s/it][A
	Running generate_until requests: 46%\|████▋ \| 116/250 [02:40<02:58, 1.33s/it][A
	Running generate_until requests: 47%\|████▋ \| 117/250 [02:41<02:56, 1.33s/it][A
	Running generate_until requests: 47%\|████▋ \| 118/250 [02:43<02:55, 1.33s/it][A
	Running generate_until requests: 48%\|████▊ \| 119/250 [02:44<02:54, 1.33s/it][A
	Running generate_until requests: 48%\|████▊ \| 120/250 [02:45<02:53, 1.33s/it][A
	Running generate_until requests: 48%\|████▊ \| 121/250 [02:47<02:51, 1.33s/it][A
	Running generate_until requests: 49%\|████▉ \| 122/250 [02:48<02:50, 1.33s/it][A
	Running generate_until requests: 49%\|████▉ \| 123/250 [02:49<02:48, 1.33s/it][A
	Running generate_until requests: 50%\|████▉ \| 124/250 [02:51<02:47, 1.33s/it][A
	Running generate_until requests: 50%\|█████ \| 125/250 [02:52<02:46, 1.33s/it][A
	Running generate_until requests: 50%\|█████ \| 126/250 [02:53<02:44, 1.33s/it][A
	Running generate_until requests: 51%\|█████ \| 127/250 [02:55<02:43, 1.33s/it][A
	Running generate_until requests: 51%\|█████ \| 128/250 [02:56<02:42, 1.33s/it][A
	Running generate_until requests: 52%\|█████▏ \| 129/250 [02:57<02:41, 1.33s/it][A
	Running generate_until requests: 52%\|█████▏ \| 130/250 [02:59<02:39, 1.33s/it][A
	Running generate_until requests: 52%\|█████▏ \| 131/250 [03:00<02:38, 1.33s/it][A
	Running generate_until requests: 53%\|█████▎ \| 132/250 [03:01<02:36, 1.33s/it][A
	Running generate_until requests: 53%\|█████▎ \| 133/250 [03:03<02:35, 1.33s/it][A
	Running generate_until requests: 54%\|█████▎ \| 134/250 [03:04<02:34, 1.33s/it][A
	Running generate_until requests: 54%\|█████▍ \| 135/250 [03:05<02:32, 1.33s/it][A
	Running generate_until requests: 54%\|█████▍ \| 136/250 [03:07<02:30, 1.32s/it][A
	Running generate_until requests: 55%\|█████▍ \| 137/250 [03:08<02:29, 1.32s/it][A
	Running generate_until requests: 55%\|█████▌ \| 138/250 [03:09<02:28, 1.32s/it][A
	Running generate_until requests: 56%\|█████▌ \| 139/250 [03:11<02:26, 1.32s/it][A
	Running generate_until requests: 56%\|█████▌ \| 140/250 [03:12<02:25, 1.33s/it][A
	Running generate_until requests: 56%\|█████▋ \| 141/250 [03:13<02:24, 1.32s/it][A
	Running generate_until requests: 57%\|█████▋ \| 142/250 [03:15<02:22, 1.32s/it][A
	Running generate_until requests: 57%\|█████▋ \| 143/250 [03:16<02:21, 1.33s/it][A
	Running generate_until requests: 58%\|█████▊ \| 144/250 [03:17<02:20, 1.33s/it][A
	Running generate_until requests: 58%\|█████▊ \| 145/250 [03:19<02:19, 1.32s/it][A
	Running generate_until requests: 58%\|█████▊ \| 146/250 [03:20<02:17, 1.33s/it][A
	Running generate_until requests: 59%\|█████▉ \| 147/250 [03:21<02:16, 1.33s/it][A
	Running generate_until requests: 59%\|█████▉ \| 148/250 [03:22<02:15, 1.33s/it][A
	Running generate_until requests: 60%\|█████▉ \| 149/250 [03:24<02:13, 1.33s/it][A
	Running generate_until requests: 60%\|██████ \| 150/250 [03:25<02:12, 1.33s/it][A
	Running generate_until requests: 60%\|██████ \| 151/250 [03:26<02:11, 1.33s/it][A
	Running generate_until requests: 61%\|██████ \| 152/250 [03:28<02:09, 1.33s/it][A
	Running generate_until requests: 61%\|██████ \| 153/250 [03:29<02:08, 1.33s/it][A
	Running generate_until requests: 62%\|██████▏ \| 154/250 [03:30<02:07, 1.32s/it][A
	Running generate_until requests: 62%\|██████▏ \| 155/250 [03:32<02:05, 1.32s/it][A
	Running generate_until requests: 62%\|██████▏ \| 156/250 [03:33<02:04, 1.32s/it][A
	Running generate_until requests: 63%\|██████▎ \| 157/250 [03:34<02:02, 1.32s/it][A
	Running generate_until requests: 63%\|██████▎ \| 158/250 [03:36<02:01, 1.32s/it][A
	Running generate_until requests: 64%\|██████▎ \| 159/250 [03:37<02:00, 1.32s/it][A
	Running generate_until requests: 64%\|██████▍ \| 160/250 [03:38<01:59, 1.32s/it][A
	Running generate_until requests: 64%\|██████▍ \| 161/250 [03:40<01:57, 1.32s/it][A
	Running generate_until requests: 65%\|██████▍ \| 162/250 [03:41<01:56, 1.32s/it][A
	Running generate_until requests: 65%\|██████▌ \| 163/250 [03:42<01:55, 1.32s/it][A
	Running generate_until requests: 66%\|██████▌ \| 164/250 [03:44<01:53, 1.32s/it][A
	Running generate_until requests: 66%\|██████▌ \| 165/250 [03:45<01:52, 1.32s/it][A
	Running generate_until requests: 66%\|██████▋ \| 166/250 [03:46<01:51, 1.32s/it][A
	Running generate_until requests: 67%\|██████▋ \| 167/250 [03:48<01:49, 1.32s/it][A
	Running generate_until requests: 67%\|██████▋ \| 168/250 [03:49<01:48, 1.32s/it][A
	Running generate_until requests: 68%\|██████▊ \| 169/250 [03:50<01:47, 1.32s/it][A
	Running generate_until requests: 68%\|██████▊ \| 170/250 [03:52<01:45, 1.32s/it][A
	Running generate_until requests: 68%\|██████▊ \| 171/250 [03:53<01:44, 1.32s/it][A
	Running generate_until requests: 69%\|██████▉ \| 172/250 [03:54<01:43, 1.32s/it][A
	Running generate_until requests: 69%\|██████▉ \| 173/250 [03:56<01:41, 1.32s/it][A
	Running generate_until requests: 70%\|██████▉ \| 174/250 [03:57<01:40, 1.32s/it][A
	Running generate_until requests: 70%\|███████ \| 175/250 [03:58<01:39, 1.32s/it][A
	Running generate_until requests: 70%\|███████ \| 176/250 [04:00<01:37, 1.32s/it][A
	Running generate_until requests: 71%\|███████ \| 177/250 [04:01<01:36, 1.32s/it][A
	Running generate_until requests: 71%\|███████ \| 178/250 [04:02<01:35, 1.32s/it][A
	Running generate_until requests: 72%\|███████▏ \| 179/250 [04:04<01:33, 1.32s/it][A
	Running generate_until requests: 72%\|███████▏ \| 180/250 [04:05<01:32, 1.32s/it][A
	Running generate_until requests: 72%\|███████▏ \| 181/250 [04:06<01:31, 1.32s/it][A
	Running generate_until requests: 73%\|███████▎ \| 182/250 [04:07<01:29, 1.32s/it][A
	Running generate_until requests: 73%\|███████▎ \| 183/250 [04:09<01:28, 1.32s/it][A
	Running generate_until requests: 74%\|███████▎ \| 184/250 [04:10<01:27, 1.32s/it][A
	Running generate_until requests: 74%\|███████▍ \| 185/250 [04:11<01:25, 1.32s/it][A
	Running generate_until requests: 74%\|███████▍ \| 186/250 [04:13<01:24, 1.32s/it][A
	Running generate_until requests: 75%\|███████▍ \| 187/250 [04:14<01:23, 1.32s/it][A
	Running generate_until requests: 75%\|███████▌ \| 188/250 [04:15<01:21, 1.32s/it][A
	Running generate_until requests: 76%\|███████▌ \| 189/250 [04:17<01:20, 1.32s/it][A
	Running generate_until requests: 76%\|███████▌ \| 190/250 [04:18<01:19, 1.32s/it][A
	Running generate_until requests: 76%\|███████▋ \| 191/250 [04:19<01:17, 1.32s/it][A
	Running generate_until requests: 77%\|███████▋ \| 192/250 [04:21<01:16, 1.32s/it][A
	Running generate_until requests: 77%\|███████▋ \| 193/250 [04:22<01:15, 1.32s/it][A
	Running generate_until requests: 78%\|███████▊ \| 194/250 [04:23<01:13, 1.32s/it][A
	Running generate_until requests: 78%\|███████▊ \| 195/250 [04:25<01:12, 1.32s/it][A
	Running generate_until requests: 78%\|███████▊ \| 196/250 [04:26<01:11, 1.32s/it][A
	Running generate_until requests: 79%\|███████▉ \| 197/250 [04:27<01:09, 1.32s/it][A
	Running generate_until requests: 79%\|███████▉ \| 198/250 [04:29<01:08, 1.32s/it][A
	Running generate_until requests: 80%\|███████▉ \| 199/250 [04:30<01:07, 1.32s/it][A
	Running generate_until requests: 80%\|████████ \| 200/250 [04:31<01:05, 1.32s/it][A
	Running generate_until requests: 80%\|████████ \| 201/250 [04:32<01:04, 1.31s/it][A
	Running generate_until requests: 81%\|████████ \| 202/250 [04:34<01:03, 1.31s/it][A
	Running generate_until requests: 81%\|████████ \| 203/250 [04:35<01:01, 1.31s/it][A
	Running generate_until requests: 82%\|████████▏ \| 204/250 [04:36<01:00, 1.31s/it][A
	Running generate_until requests: 82%\|████████▏ \| 205/250 [04:38<00:56, 1.25s/it][A
	Running generate_until requests: 82%\|████████▏ \| 206/250 [04:39<00:53, 1.21s/it][A
	Running generate_until requests: 83%\|████████▎ \| 207/250 [04:40<00:50, 1.18s/it][A
	Running generate_until requests: 83%\|████████▎ \| 208/250 [04:41<00:48, 1.16s/it][A
	Running generate_until requests: 84%\|████████▎ \| 209/250 [04:42<00:46, 1.14s/it][A
	Running generate_until requests: 84%\|████████▍ \| 210/250 [04:43<00:45, 1.13s/it][A
	Running generate_until requests: 84%\|████████▍ \| 211/250 [04:44<00:43, 1.12s/it][A
	Running generate_until requests: 85%\|████████▍ \| 212/250 [04:45<00:42, 1.12s/it][A
	Running generate_until requests: 85%\|████████▌ \| 213/250 [04:46<00:41, 1.12s/it][A
	Running generate_until requests: 86%\|████████▌ \| 214/250 [04:48<00:40, 1.11s/it][A
	Running generate_until requests: 86%\|████████▌ \| 215/250 [04:49<00:38, 1.11s/it][A
	Running generate_until requests: 86%\|████████▋ \| 216/250 [04:50<00:37, 1.11s/it][A
	Running generate_until requests: 87%\|████████▋ \| 217/250 [04:51<00:36, 1.11s/it][A
	Running generate_until requests: 87%\|████████▋ \| 218/250 [04:52<00:35, 1.11s/it][A
	Running generate_until requests: 88%\|████████▊ \| 219/250 [04:53<00:34, 1.10s/it][A
	Running generate_until requests: 88%\|████████▊ \| 220/250 [04:54<00:33, 1.10s/it][A
	Running generate_until requests: 88%\|████████▊ \| 221/250 [04:55<00:31, 1.10s/it][A
	Running generate_until requests: 89%\|████████▉ \| 222/250 [04:56<00:30, 1.10s/it][A
	Running generate_until requests: 89%\|████████▉ \| 223/250 [04:57<00:29, 1.10s/it][A
	Running generate_until requests: 90%\|████████▉ \| 224/250 [04:59<00:28, 1.10s/it][A
	Running generate_until requests: 90%\|█████████ \| 225/250 [05:00<00:27, 1.10s/it][A
	Running generate_until requests: 90%\|█████████ \| 226/250 [05:01<00:26, 1.10s/it][A
	Running generate_until requests: 91%\|█████████ \| 227/250 [05:02<00:25, 1.10s/it][A
	Running generate_until requests: 91%\|█████████ \| 228/250 [05:03<00:24, 1.09s/it][A
	Running generate_until requests: 92%\|█████████▏\| 229/250 [05:04<00:22, 1.09s/it][A
	Running generate_until requests: 92%\|█████████▏\| 230/250 [05:05<00:21, 1.09s/it][A
	Running generate_until requests: 92%\|█████████▏\| 231/250 [05:06<00:20, 1.09s/it][A
	Running generate_until requests: 93%\|█████████▎\| 232/250 [05:07<00:19, 1.08s/it][A
	Running generate_until requests: 93%\|█████████▎\| 233/250 [05:08<00:18, 1.08s/it][A
	Running generate_until requests: 94%\|█████████▎\| 234/250 [05:09<00:17, 1.08s/it][A
	Running generate_until requests: 94%\|█████████▍\| 235/250 [05:10<00:16, 1.08s/it][A
	Running generate_until requests: 94%\|█████████▍\| 236/250 [05:12<00:15, 1.07s/it][A
	Running generate_until requests: 95%\|█████████▍\| 237/250 [05:13<00:13, 1.07s/it][A
	Running generate_until requests: 95%\|█████████▌\| 238/250 [05:14<00:12, 1.06s/it][A
	Running generate_until requests: 96%\|█████████▌\| 239/250 [05:15<00:11, 1.06s/it][A
	Running generate_until requests: 96%\|█████████▌\| 240/250 [05:16<00:10, 1.06s/it][A
	Running generate_until requests: 96%\|█████████▋\| 241/250 [05:17<00:09, 1.05s/it][A
	Running generate_until requests: 97%\|█████████▋\| 242/250 [05:18<00:08, 1.05s/it][A
	Running generate_until requests: 97%\|█████████▋\| 243/250 [05:19<00:07, 1.05s/it][A
	Running generate_until requests: 98%\|█████████▊\| 244/250 [05:20<00:06, 1.05s/it][A
	Running generate_until requests: 98%\|█████████▊\| 245/250 [05:21<00:05, 1.05s/it][A
	Running generate_until requests: 98%\|█████████▊\| 246/250 [05:22<00:04, 1.04s/it][A
	Running generate_until requests: 99%\|█████████▉\| 248/250 [05:24<00:02, 1.01s/it][A
	Running generate_until requests: 100%\|██████████\| 250/250 [05:26<00:00, 1.01it/s][A Running generate_until requests: 100%\|██████████\| 250/250 [05:26<00:00, 1.31s/it]
	2026-07-18:15:23:38 INFO [loggers.evaluation_tracker:247] Saving results aggregated
	2026-07-18:15:23:38 INFO [loggers.evaluation_tracker:119] Saving per-task samples to /workspace/outputs/eval/token_t045/babilong8k_qa1_qa5_n50/__workspace__outputs__l2a_style__stage2__checkpoint-25/*.jsonl
	hf ({'pretrained': '/workspace/outputs/l2a_style/stage2/checkpoint-25', 'dtype': 'bfloat16', 'max_length': 16384, 'attn_implementation': 'sdpa'}), gen_kwargs: ({}), limit: 50.0, num_fewshot: 2, batch_size: 1
	\| Tasks \|Version\|Filter\|n-shot\|Metric\| \|Value\| \|Stderr\|
	\|----------------\|------:\|------\|-----:\|------\|---\|----:\|---\|-----:\|
	\|babilong_longctx\| 0\|none \| \|acc \|↑ \|0.588\|± \|0.0298\|
	\| - babilong_qa1 \| 0\|none \| 2\|acc \|↑ \|0.640\|± \|0.0686\|
	\| - babilong_qa2 \| 0\|none \| 2\|acc \|↑ \|0.480\|± \|0.0714\|
	\| - babilong_qa3 \| 0\|none \| 2\|acc \|↑ \|0.340\|± \|0.0677\|
	\| - babilong_qa4 \| 0\|none \| 2\|acc \|↑ \|0.720\|± \|0.0641\|
	\| - babilong_qa5 \| 0\|none \| 2\|acc \|↑ \|0.760\|± \|0.0610\|

	\| Groups \|Version\|Filter\|n-shot\|Metric\| \|Value\| \|Stderr\|
	\|----------------\|------:\|------\|------\|------\|---\|----:\|---\|-----:\|
	\|babilong_longctx\| 0\|none \| \|acc \|↑ \|0.588\|± \|0.0298\|