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 /ruler_a.log

keepsloading

Upload folder using huggingface_hub

6a26953 verified about 1 month ago

Raw

History Blame Contribute Delete

32 kB

	2026-07-18:14:53:21 WARNING [config.evaluate_config:281] --limit SHOULD ONLY BE USED FOR TESTING. REAL METRICS SHOULD NOT BE COMPUTED USING LIMIT.
	2026-07-18:14:53:25 INFO [_cli.run:376] Selected Tasks: ['niah_single_1', 'niah_single_3', 'niah_multikey_2', 'niah_multiquery', 'ruler_vt', 'ruler_fwe', 'ruler_qa_hotpot']
	2026-07-18:14:53:26 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
	2026-07-18:14:53:26 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'}
	2026-07-18:14:53:29 INFO [models.huggingface:161] Using device 'cuda:0'
	2026-07-18:14:53:29 INFO [models.huggingface:548] Model type cannot be determined. Using default model type 'causal'
	2026-07-18:14:53:29 INFO [models.huggingface:423] Model parallel was set to False, max memory was not set, and device map was set to {'': 'cuda:0'}
	2026-07-18:14:53:38 WARNING [api.task:856] niah_single_1: 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.
	2026-07-18:14:53:38 INFO [tasks.ruler.common_utils:26] Using tokenizer /workspace/outputs/l2a_style/stage2/checkpoint-25 for synthetic tasks.

	Generating synthetic samples: repeat \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating synthetic samples: repeat \| 8192: 12%\|█▏ \| 60/500 [00:01<00:07, 59.82it/s][A
	Generating synthetic samples: repeat \| 8192: 24%\|██▍ \| 120/500 [00:02<00:06, 59.14it/s][A
	Generating synthetic samples: repeat \| 8192: 36%\|███▌ \| 180/500 [00:03<00:05, 59.14it/s][A
	Generating synthetic samples: repeat \| 8192: 48%\|████▊ \| 240/500 [00:04<00:04, 58.38it/s][A
	Generating synthetic samples: repeat \| 8192: 60%\|█████▉ \| 299/500 [00:05<00:03, 58.25it/s][A
	Generating synthetic samples: repeat \| 8192: 72%\|███████▏ \| 360/500 [00:06<00:02, 58.88it/s][A
	Generating synthetic samples: repeat \| 8192: 84%\|████████▍ \| 420/500 [00:07<00:01, 59.11it/s][A
	Generating synthetic samples: repeat \| 8192: 96%\|█████████▌\| 481/500 [00:08<00:00, 59.62it/s][A Generating synthetic samples: repeat \| 8192: 100%\|██████████\| 500/500 [00:08<00:00, 59.07it/s]
	2026-07-18:14:53:47 WARNING [api.task:856] niah_single_3: 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.

	Generating synthetic samples: essay \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating synthetic samples: essay \| 8192: 11%\|█ \| 53/500 [00:01<00:08, 53.00it/s][A
	Generating synthetic samples: essay \| 8192: 22%\|██▏ \| 109/500 [00:02<00:07, 54.23it/s][A
	Generating synthetic samples: essay \| 8192: 33%\|███▎ \| 166/500 [00:03<00:06, 55.44it/s][A
	Generating synthetic samples: essay \| 8192: 44%\|████▍ \| 222/500 [00:04<00:05, 55.26it/s][A
	Generating synthetic samples: essay \| 8192: 56%\|█████▌ \| 278/500 [00:05<00:04, 55.05it/s][A
	Generating synthetic samples: essay \| 8192: 67%\|██████▋ \| 335/500 [00:06<00:02, 55.63it/s][A
	Generating synthetic samples: essay \| 8192: 78%\|███████▊ \| 391/500 [00:07<00:01, 55.67it/s][A
	Generating synthetic samples: essay \| 8192: 90%\|████████▉ \| 448/500 [00:08<00:00, 56.00it/s][A Generating synthetic samples: essay \| 8192: 100%\|██████████\| 500/500 [00:09<00:00, 55.55it/s]
	2026-07-18:14:53:58 WARNING [api.task:856] niah_multikey_2: 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.

	Generating synthetic samples: needle \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating synthetic samples: needle \| 8192: 13%\|█▎ \| 66/500 [00:01<00:06, 65.25it/s][A
	Generating synthetic samples: needle \| 8192: 27%\|██▋ \| 134/500 [00:02<00:05, 66.66it/s][A
	Generating synthetic samples: needle \| 8192: 41%\|████ \| 203/500 [00:03<00:04, 67.30it/s][A
	Generating synthetic samples: needle \| 8192: 54%\|█████▍ \| 272/500 [00:04<00:03, 67.92it/s][A
	Generating synthetic samples: needle \| 8192: 68%\|██████▊ \| 340/500 [00:05<00:02, 67.36it/s][A
	Generating synthetic samples: needle \| 8192: 82%\|████████▏ \| 408/500 [00:06<00:01, 66.26it/s][A
	Generating synthetic samples: needle \| 8192: 95%\|█████████▌\| 477/500 [00:07<00:00, 66.96it/s][A Generating synthetic samples: needle \| 8192: 100%\|██████████\| 500/500 [00:07<00:00, 66.98it/s]
	2026-07-18:14:54:05 WARNING [api.task:856] niah_multiquery: 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.

	Generating synthetic samples: essay \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating synthetic samples: essay \| 8192: 11%\|█ \| 54/500 [00:01<00:08, 53.65it/s][A
	Generating synthetic samples: essay \| 8192: 22%\|██▏ \| 109/500 [00:02<00:07, 54.13it/s][A
	Generating synthetic samples: essay \| 8192: 33%\|███▎ \| 164/500 [00:03<00:06, 54.13it/s][A
	Generating synthetic samples: essay \| 8192: 44%\|████▍ \| 219/500 [00:04<00:05, 53.66it/s][A
	Generating synthetic samples: essay \| 8192: 55%\|█████▍ \| 273/500 [00:05<00:04, 53.51it/s][A
	Generating synthetic samples: essay \| 8192: 66%\|██████▌ \| 329/500 [00:06<00:03, 54.13it/s][A
	Generating synthetic samples: essay \| 8192: 77%\|███████▋ \| 384/500 [00:07<00:02, 53.52it/s][A
	Generating synthetic samples: essay \| 8192: 88%\|████████▊ \| 438/500 [00:08<00:01, 53.27it/s][A
	Generating synthetic samples: essay \| 8192: 98%\|█████████▊\| 492/500 [00:09<00:00, 53.46it/s][A Generating synthetic samples: essay \| 8192: 100%\|██████████\| 500/500 [00:09<00:00, 53.65it/s]
	2026-07-18:14:54:15 WARNING [api.task:856] ruler_vt: 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.
	2026-07-18:14:54:15 INFO [tasks.ruler.common_utils:26] Using tokenizer /workspace/outputs/l2a_style/stage2/checkpoint-25 for synthetic tasks.
	Max length 500 \| Current length 307 \| Noises: 5
	Max length 500 \| Current length 410 \| Noises: 10
	Max length 500 \| Current length 531 \| Noises: 15
	Num noises: 10

	0%\| \| 0/1 [00:00<?, ?it/s][A 100%\|██████████\| 1/1 [00:00<00:00, 949.80it/s]
	Max length 8192 \| Current length 791 \| Noises: 10
	Max length 8192 \| Current length 1025 \| Noises: 20
	Max length 8192 \| Current length 1273 \| Noises: 30
	Max length 8192 \| Current length 1510 \| Noises: 40
	Max length 8192 \| Current length 1749 \| Noises: 50
	Max length 8192 \| Current length 1996 \| Noises: 60
	Max length 8192 \| Current length 2231 \| Noises: 70
	Max length 8192 \| Current length 2473 \| Noises: 80
	Max length 8192 \| Current length 2705 \| Noises: 90
	Max length 8192 \| Current length 2953 \| Noises: 100
	Max length 8192 \| Current length 3186 \| Noises: 110
	Max length 8192 \| Current length 3431 \| Noises: 120
	Max length 8192 \| Current length 3671 \| Noises: 130
	Max length 8192 \| Current length 3916 \| Noises: 140
	Max length 8192 \| Current length 4149 \| Noises: 150
	Max length 8192 \| Current length 4386 \| Noises: 160
	Max length 8192 \| Current length 4630 \| Noises: 170
	Max length 8192 \| Current length 4871 \| Noises: 180
	Max length 8192 \| Current length 5110 \| Noises: 190
	Max length 8192 \| Current length 5349 \| Noises: 200
	Max length 8192 \| Current length 5590 \| Noises: 210
	Max length 8192 \| Current length 5832 \| Noises: 220
	Max length 8192 \| Current length 6070 \| Noises: 230
	Max length 8192 \| Current length 6311 \| Noises: 240
	Max length 8192 \| Current length 6544 \| Noises: 250
	Max length 8192 \| Current length 6791 \| Noises: 260
	Max length 8192 \| Current length 7031 \| Noises: 270
	Max length 8192 \| Current length 7271 \| Noises: 280
	Max length 8192 \| Current length 7513 \| Noises: 290
	Max length 8192 \| Current length 7751 \| Noises: 300
	Max length 8192 \| Current length 7986 \| Noises: 310
	Max length 8192 \| Current length 8230 \| Noises: 320
	Num noises: 310

	0%\| \| 0/500 [00:00<?, ?it/s][A
	12%\|█▏ \| 58/500 [00:01<00:07, 57.66it/s][A
	23%\|██▎ \| 116/500 [00:02<00:06, 57.08it/s][A
	35%\|███▍ \| 174/500 [00:03<00:05, 57.31it/s][A
	46%\|████▋ \| 232/500 [00:04<00:04, 57.12it/s][A
	58%\|█████▊ \| 290/500 [00:05<00:03, 57.13it/s][A
	70%\|██████▉ \| 348/500 [00:06<00:02, 57.22it/s][A
	81%\|████████ \| 406/500 [00:07<00:01, 57.00it/s][A
	93%\|█████████▎\| 464/500 [00:08<00:00, 56.37it/s][A 100%\|██████████\| 500/500 [00:08<00:00, 56.81it/s]
	2026-07-18:14:54:25 WARNING [api.task:856] ruler_fwe: 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.

	Generating FWE Samples \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating FWE Samples \| 8192: 8%\|▊ \| 40/500 [00:01<00:11, 39.40it/s][A
	Generating FWE Samples \| 8192: 16%\|█▌ \| 80/500 [00:02<00:10, 39.69it/s][A
	Generating FWE Samples \| 8192: 24%\|██▍ \| 120/500 [00:03<00:09, 39.55it/s][A
	Generating FWE Samples \| 8192: 32%\|███▏ \| 160/500 [00:04<00:08, 39.59it/s][A
	Generating FWE Samples \| 8192: 40%\|████ \| 200/500 [00:05<00:07, 38.93it/s][A
	Generating FWE Samples \| 8192: 48%\|████▊ \| 239/500 [00:06<00:06, 37.70it/s][A
	Generating FWE Samples \| 8192: 55%\|█████▌ \| 277/500 [00:07<00:05, 37.28it/s][A
	Generating FWE Samples \| 8192: 63%\|██████▎ \| 315/500 [00:08<00:05, 36.93it/s][A
	Generating FWE Samples \| 8192: 70%\|███████ \| 352/500 [00:09<00:04, 36.92it/s][A
	Generating FWE Samples \| 8192: 78%\|███████▊ \| 389/500 [00:10<00:03, 36.70it/s][A
	Generating FWE Samples \| 8192: 85%\|████████▌ \| 427/500 [00:11<00:01, 36.99it/s][A
	Generating FWE Samples \| 8192: 93%\|█████████▎\| 464/500 [00:12<00:00, 36.81it/s][A Generating FWE Samples \| 8192: 100%\|██████████\| 500/500 [00:13<00:00, 37.48it/s]
	2026-07-18:14:54:38 WARNING [api.task:856] ruler_qa_hotpot: 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.

	Generating QA Samples \| 8192: 0%\| \| 0/500 [00:00<?, ?it/s][A
	Generating QA Samples \| 8192: 5%\|▍ \| 24/500 [00:01<00:20, 23.41it/s][A
	Generating QA Samples \| 8192: 10%\|▉ \| 49/500 [00:02<00:18, 23.76it/s][A
	Generating QA Samples \| 8192: 15%\|█▍ \| 73/500 [00:03<00:17, 23.74it/s][A
	Generating QA Samples \| 8192: 20%\|█▉ \| 98/500 [00:04<00:16, 23.96it/s][A
	Generating QA Samples \| 8192: 25%\|██▌ \| 126/500 [00:05<00:14, 25.13it/s][A
	Generating QA Samples \| 8192: 30%\|███ \| 152/500 [00:06<00:13, 25.20it/s][A
	Generating QA Samples \| 8192: 36%\|███▌ \| 178/500 [00:07<00:12, 25.26it/s][A
	Generating QA Samples \| 8192: 41%\|████ \| 204/500 [00:08<00:11, 24.68it/s][A
	Generating QA Samples \| 8192: 46%\|████▌ \| 229/500 [00:09<00:10, 24.69it/s][A
	Generating QA Samples \| 8192: 51%\|█████ \| 254/500 [00:10<00:10, 24.53it/s][A
	Generating QA Samples \| 8192: 56%\|█████▌ \| 279/500 [00:11<00:08, 24.64it/s][A
	Generating QA Samples \| 8192: 61%\|██████ \| 306/500 [00:12<00:07, 25.12it/s][A
	Generating QA Samples \| 8192: 66%\|██████▋ \| 332/500 [00:13<00:06, 24.75it/s][A
	Generating QA Samples \| 8192: 72%\|███████▏ \| 358/500 [00:14<00:05, 24.88it/s][A
	Generating QA Samples \| 8192: 77%\|███████▋ \| 384/500 [00:15<00:04, 25.02it/s][A
	Generating QA Samples \| 8192: 82%\|████████▏ \| 410/500 [00:16<00:03, 25.01it/s][A
	Generating QA Samples \| 8192: 87%\|████████▋ \| 436/500 [00:17<00:02, 25.17it/s][A
	Generating QA Samples \| 8192: 92%\|█████████▏\| 462/500 [00:18<00:01, 24.90it/s][A
	Generating QA Samples \| 8192: 98%\|█████████▊\| 489/500 [00:19<00:00, 25.39it/s][A Generating QA Samples \| 8192: 100%\|██████████\| 500/500 [00:20<00:00, 24.86it/s]
	2026-07-18:14:55:01 INFO [tasks:700] Selected tasks:
	2026-07-18:14:55:01 INFO [tasks:691] Task: ruler_qa_hotpot (ruler/qa_hotpot.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: ruler_fwe (ruler/fwe.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: ruler_vt (ruler/vt.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: niah_multiquery (ruler/niah_multiquery.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: niah_multikey_2 (ruler/niah_multikey_2.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: niah_single_3 (ruler/niah_single_3.yaml)
	2026-07-18:14:55:01 INFO [tasks:691] Task: niah_single_1 (ruler/niah_single_1.yaml)
	2026-07-18:14:55:01 INFO [evaluator:314] ruler_qa_hotpot: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 32, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] ruler_fwe: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 50, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] ruler_vt: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 30, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] niah_multiquery: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 128, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] niah_multikey_2: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 128, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] niah_single_3: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 128, 'until': []}
	2026-07-18:14:55:01 INFO [evaluator:314] niah_single_1: Using gen_kwargs: {'do_sample': False, 'temperature': 0.0, 'max_gen_toks': 128, 'until': []}
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for ruler_qa_hotpot on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1017.21it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for ruler_fwe on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 982.40it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for ruler_vt on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1110.71it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for niah_multiquery on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1076.43it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for niah_multikey_2 on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1155.15it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for niah_single_3 on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1095.16it/s]
	2026-07-18:14:55:01 INFO [api.task:311] Building contexts for niah_single_1 on rank 0...

	0%\| \| 0/20 [00:00<?, ?it/s][A 100%\|██████████\| 20/20 [00:00<00:00, 1130.34it/s]
	2026-07-18:14:55:01 INFO [evaluator:584] Running generate_until requests

	Running generate_until requests: 0%\| \| 0/140 [00:00<?, ?it/s][A
	Running generate_until requests: 1%\| \| 1/140 [00:03<07:30, 3.24s/it][A
	Running generate_until requests: 1%\|▏ \| 2/140 [00:05<06:09, 2.68s/it][A
	Running generate_until requests: 2%\|▏ \| 3/140 [00:07<05:35, 2.45s/it][A
	Running generate_until requests: 3%\|▎ \| 4/140 [00:09<05:04, 2.24s/it][A
	Running generate_until requests: 4%\|▎ \| 5/140 [00:11<04:47, 2.13s/it][A
	Running generate_until requests: 4%\|▍ \| 6/140 [00:13<04:35, 2.06s/it][A
	Running generate_until requests: 5%\|▌ \| 7/140 [00:15<04:25, 2.00s/it][A
	Running generate_until requests: 6%\|▌ \| 8/140 [00:17<04:18, 1.96s/it][A
	Running generate_until requests: 6%\|▋ \| 9/140 [00:19<04:12, 1.93s/it][A
	Running generate_until requests: 7%\|▋ \| 10/140 [00:20<04:06, 1.90s/it][A
	Running generate_until requests: 8%\|▊ \| 11/140 [00:22<04:02, 1.88s/it][A
	Running generate_until requests: 9%\|▊ \| 12/140 [00:24<03:58, 1.86s/it][A
	Running generate_until requests: 9%\|▉ \| 13/140 [00:26<03:54, 1.85s/it][A
	Running generate_until requests: 10%\|█ \| 14/140 [00:28<03:50, 1.83s/it][A
	Running generate_until requests: 11%\|█ \| 15/140 [00:29<03:48, 1.82s/it][A
	Running generate_until requests: 11%\|█▏ \| 16/140 [00:31<03:44, 1.81s/it][A
	Running generate_until requests: 12%\|█▏ \| 17/140 [00:33<03:41, 1.80s/it][A
	Running generate_until requests: 13%\|█▎ \| 18/140 [00:35<03:37, 1.79s/it][A
	Running generate_until requests: 14%\|█▎ \| 19/140 [00:37<03:34, 1.78s/it][A
	Running generate_until requests: 14%\|█▍ \| 20/140 [00:38<03:31, 1.76s/it][A
	Running generate_until requests: 15%\|█▌ \| 21/140 [00:41<04:13, 2.13s/it][A
	Running generate_until requests: 16%\|█▌ \| 22/140 [00:44<04:32, 2.31s/it][A
	Running generate_until requests: 16%\|█▋ \| 23/140 [00:47<04:44, 2.44s/it][A
	Running generate_until requests: 17%\|█▋ \| 24/140 [00:49<04:52, 2.52s/it][A
	Running generate_until requests: 18%\|█▊ \| 25/140 [00:52<04:56, 2.58s/it][A
	Running generate_until requests: 19%\|█▊ \| 26/140 [00:55<04:57, 2.61s/it][A
	Running generate_until requests: 19%\|█▉ \| 27/140 [00:58<04:57, 2.64s/it][A
	Running generate_until requests: 20%\|██ \| 28/140 [01:00<04:57, 2.65s/it][A
	Running generate_until requests: 21%\|██ \| 29/140 [01:03<04:57, 2.68s/it][A
	Running generate_until requests: 21%\|██▏ \| 30/140 [01:06<04:54, 2.68s/it][A
	Running generate_until requests: 22%\|██▏ \| 31/140 [01:08<04:52, 2.68s/it][A
	Running generate_until requests: 23%\|██▎ \| 32/140 [01:11<04:49, 2.68s/it][A
	Running generate_until requests: 24%\|██▎ \| 33/140 [01:14<04:47, 2.68s/it][A
	Running generate_until requests: 24%\|██▍ \| 34/140 [01:16<04:43, 2.68s/it][A
	Running generate_until requests: 25%\|██▌ \| 35/140 [01:19<04:40, 2.68s/it][A
	Running generate_until requests: 26%\|██▌ \| 36/140 [01:22<04:38, 2.67s/it][A
	Running generate_until requests: 26%\|██▋ \| 37/140 [01:24<04:35, 2.67s/it][A
	Running generate_until requests: 27%\|██▋ \| 38/140 [01:27<04:32, 2.67s/it][A
	Running generate_until requests: 28%\|██▊ \| 39/140 [01:30<04:28, 2.66s/it][A
	Running generate_until requests: 29%\|██▊ \| 40/140 [01:32<04:26, 2.66s/it][A
	Running generate_until requests: 29%\|██▉ \| 41/140 [01:35<04:08, 2.51s/it][A
	Running generate_until requests: 30%\|███ \| 42/140 [01:36<03:45, 2.30s/it][A
	Running generate_until requests: 31%\|███ \| 43/140 [01:38<03:29, 2.15s/it][A
	Running generate_until requests: 31%\|███▏ \| 44/140 [01:40<03:17, 2.05s/it][A
	Running generate_until requests: 32%\|███▏ \| 45/140 [01:42<03:08, 1.98s/it][A
	Running generate_until requests: 33%\|███▎ \| 46/140 [01:44<03:01, 1.93s/it][A
	Running generate_until requests: 34%\|███▎ \| 47/140 [01:45<02:56, 1.90s/it][A
	Running generate_until requests: 34%\|███▍ \| 48/140 [01:47<02:52, 1.87s/it][A
	Running generate_until requests: 35%\|███▌ \| 49/140 [01:49<02:48, 1.85s/it][A
	Running generate_until requests: 36%\|███▌ \| 50/140 [01:51<02:45, 1.84s/it][A
	Running generate_until requests: 36%\|███▋ \| 51/140 [01:53<02:43, 1.83s/it][A
	Running generate_until requests: 37%\|███▋ \| 52/140 [01:54<02:40, 1.82s/it][A
	Running generate_until requests: 38%\|███▊ \| 53/140 [01:56<02:38, 1.82s/it][A
	Running generate_until requests: 39%\|███▊ \| 54/140 [01:58<02:36, 1.82s/it][A
	Running generate_until requests: 39%\|███▉ \| 55/140 [02:00<02:34, 1.81s/it][A
	Running generate_until requests: 40%\|████ \| 56/140 [02:02<02:32, 1.81s/it][A
	Running generate_until requests: 41%\|████ \| 57/140 [02:03<02:30, 1.81s/it][A
	Running generate_until requests: 41%\|████▏ \| 58/140 [02:05<02:28, 1.81s/it][A
	Running generate_until requests: 42%\|████▏ \| 59/140 [02:07<02:26, 1.81s/it][A
	Running generate_until requests: 43%\|████▎ \| 60/140 [02:09<02:24, 1.81s/it][A
	Running generate_until requests: 44%\|████▎ \| 61/140 [02:17<04:41, 3.56s/it][A
	Running generate_until requests: 44%\|████▍ \| 62/140 [02:23<05:42, 4.40s/it][A
	Running generate_until requests: 45%\|████▌ \| 63/140 [02:29<06:23, 4.98s/it][A
	Running generate_until requests: 46%\|████▌ \| 64/140 [02:36<06:48, 5.38s/it][A
	Running generate_until requests: 46%\|████▋ \| 65/140 [02:42<07:04, 5.66s/it][A
	Running generate_until requests: 47%\|████▋ \| 66/140 [02:48<07:13, 5.86s/it][A
	Running generate_until requests: 48%\|████▊ \| 67/140 [02:55<07:18, 6.00s/it][A
	Running generate_until requests: 49%\|████▊ \| 68/140 [03:01<07:18, 6.09s/it][A
	Running generate_until requests: 49%\|████▉ \| 69/140 [03:07<07:17, 6.16s/it][A
	Running generate_until requests: 50%\|█████ \| 70/140 [03:13<07:14, 6.21s/it][A
	Running generate_until requests: 51%\|█████ \| 71/140 [03:20<07:10, 6.24s/it][A
	Running generate_until requests: 51%\|█████▏ \| 72/140 [03:26<07:06, 6.27s/it][A
	Running generate_until requests: 52%\|█████▏ \| 73/140 [03:32<07:01, 6.29s/it][A
	Running generate_until requests: 53%\|█████▎ \| 74/140 [03:39<06:55, 6.29s/it][A
	Running generate_until requests: 54%\|█████▎ \| 75/140 [03:45<06:49, 6.30s/it][A
	Running generate_until requests: 54%\|█████▍ \| 76/140 [03:51<06:43, 6.31s/it][A
	Running generate_until requests: 55%\|█████▌ \| 77/140 [03:58<06:37, 6.31s/it][A
	Running generate_until requests: 56%\|█████▌ \| 78/140 [04:04<06:31, 6.31s/it][A
	Running generate_until requests: 56%\|█████▋ \| 79/140 [04:10<06:25, 6.31s/it][A
	Running generate_until requests: 57%\|█████▋ \| 80/140 [04:17<06:19, 6.32s/it][A
	Running generate_until requests: 58%\|█████▊ \| 81/140 [04:23<06:13, 6.32s/it][A
	Running generate_until requests: 59%\|█████▊ \| 82/140 [04:29<06:06, 6.32s/it][A
	Running generate_until requests: 59%\|█████▉ \| 83/140 [04:36<06:00, 6.32s/it][A
	Running generate_until requests: 60%\|██████ \| 84/140 [04:42<05:53, 6.32s/it][A
	Running generate_until requests: 61%\|██████ \| 85/140 [04:48<05:46, 6.31s/it][A
	Running generate_until requests: 61%\|██████▏ \| 86/140 [04:55<05:42, 6.34s/it][A
	Running generate_until requests: 62%\|██████▏ \| 87/140 [05:01<05:36, 6.35s/it][A
	Running generate_until requests: 63%\|██████▎ \| 88/140 [05:07<05:29, 6.34s/it][A
	Running generate_until requests: 64%\|██████▎ \| 89/140 [05:14<05:23, 6.33s/it][A
	Running generate_until requests: 64%\|██████▍ \| 90/140 [05:20<05:16, 6.33s/it][A
	Running generate_until requests: 65%\|██████▌ \| 91/140 [05:26<05:10, 6.34s/it][A
	Running generate_until requests: 66%\|██████▌ \| 92/140 [05:33<05:04, 6.33s/it][A
	Running generate_until requests: 66%\|██████▋ \| 93/140 [05:39<04:57, 6.33s/it][A
	Running generate_until requests: 67%\|██████▋ \| 94/140 [05:45<04:51, 6.34s/it][A
	Running generate_until requests: 68%\|██████▊ \| 95/140 [05:52<04:45, 6.34s/it][A
	Running generate_until requests: 69%\|██████▊ \| 96/140 [05:58<04:38, 6.33s/it][A
	Running generate_until requests: 69%\|██████▉ \| 97/140 [06:04<04:32, 6.34s/it][A
	Running generate_until requests: 70%\|███████ \| 98/140 [06:11<04:26, 6.34s/it][A
	Running generate_until requests: 71%\|███████ \| 99/140 [06:17<04:19, 6.34s/it][A
	Running generate_until requests: 71%\|███████▏ \| 100/140 [06:23<04:13, 6.34s/it][A
	Running generate_until requests: 72%\|███████▏ \| 101/140 [06:30<04:07, 6.33s/it][A
	Running generate_until requests: 73%\|███████▎ \| 102/140 [06:36<04:00, 6.33s/it][A
	Running generate_until requests: 74%\|███████▎ \| 103/140 [06:42<03:54, 6.33s/it][A
	Running generate_until requests: 74%\|███████▍ \| 104/140 [06:49<03:47, 6.32s/it][A
	Running generate_until requests: 75%\|███████▌ \| 105/140 [06:55<03:41, 6.32s/it][A
	Running generate_until requests: 76%\|███████▌ \| 106/140 [07:01<03:34, 6.32s/it][A
	Running generate_until requests: 76%\|███████▋ \| 107/140 [07:08<03:28, 6.31s/it][A
	Running generate_until requests: 77%\|███████▋ \| 108/140 [07:14<03:21, 6.31s/it][A
	Running generate_until requests: 78%\|███████▊ \| 109/140 [07:20<03:15, 6.32s/it][A
	Running generate_until requests: 79%\|███████▊ \| 110/140 [07:27<03:09, 6.33s/it][A
	Running generate_until requests: 79%\|███████▉ \| 111/140 [07:33<03:03, 6.33s/it][A
	Running generate_until requests: 80%\|████████ \| 112/140 [07:39<02:56, 6.32s/it][A
	Running generate_until requests: 81%\|████████ \| 113/140 [07:46<02:50, 6.32s/it][A
	Running generate_until requests: 81%\|████████▏ \| 114/140 [07:52<02:44, 6.33s/it][A
	Running generate_until requests: 82%\|████████▏ \| 115/140 [07:58<02:37, 6.32s/it][A
	Running generate_until requests: 83%\|████████▎ \| 116/140 [08:04<02:31, 6.32s/it][A
	Running generate_until requests: 84%\|████████▎ \| 117/140 [08:11<02:24, 6.30s/it][A
	Running generate_until requests: 84%\|████████▍ \| 118/140 [08:17<02:18, 6.30s/it][A
	Running generate_until requests: 85%\|████████▌ \| 119/140 [08:23<02:12, 6.30s/it][A
	Running generate_until requests: 86%\|████████▌ \| 120/140 [08:30<02:05, 6.29s/it][A
	Running generate_until requests: 86%\|████████▋ \| 121/140 [08:36<01:59, 6.29s/it][A
	Running generate_until requests: 87%\|████████▋ \| 122/140 [08:42<01:53, 6.29s/it][A
	Running generate_until requests: 88%\|████████▊ \| 123/140 [08:48<01:46, 6.29s/it][A
	Running generate_until requests: 89%\|████████▊ \| 124/140 [08:55<01:40, 6.28s/it][A
	Running generate_until requests: 89%\|████████▉ \| 125/140 [09:01<01:34, 6.28s/it][A
	Running generate_until requests: 90%\|█████████ \| 126/140 [09:07<01:27, 6.28s/it][A
	Running generate_until requests: 91%\|█████████ \| 127/140 [09:14<01:21, 6.28s/it][A
	Running generate_until requests: 91%\|█████████▏\| 128/140 [09:20<01:15, 6.28s/it][A
	Running generate_until requests: 92%\|█████████▏\| 129/140 [09:26<01:08, 6.27s/it][A
	Running generate_until requests: 93%\|█████████▎\| 130/140 [09:32<01:02, 6.26s/it][A
	Running generate_until requests: 94%\|█████████▎\| 131/140 [09:39<00:56, 6.26s/it][A
	Running generate_until requests: 94%\|█████████▍\| 132/140 [09:45<00:50, 6.26s/it][A
	Running generate_until requests: 95%\|█████████▌\| 133/140 [09:51<00:43, 6.26s/it][A
	Running generate_until requests: 96%\|█████████▌\| 134/140 [09:57<00:37, 6.25s/it][A
	Running generate_until requests: 96%\|█████████▋\| 135/140 [10:04<00:31, 6.25s/it][A
	Running generate_until requests: 97%\|█████████▋\| 136/140 [10:10<00:24, 6.25s/it][A
	Running generate_until requests: 98%\|█████████▊\| 137/140 [10:16<00:18, 6.25s/it][A
	Running generate_until requests: 99%\|█████████▊\| 138/140 [10:22<00:12, 6.25s/it][A
	Running generate_until requests: 99%\|█████████▉\| 139/140 [10:29<00:06, 6.25s/it][A
	Running generate_until requests: 100%\|██████████\| 140/140 [10:35<00:00, 6.25s/it][A Running generate_until requests: 100%\|██████████\| 140/140 [10:35<00:00, 4.54s/it]
	2026-07-18:15:05:41 INFO [loggers.evaluation_tracker:247] Saving results aggregated
	2026-07-18:15:05:41 INFO [loggers.evaluation_tracker:119] Saving per-task samples to /workspace/outputs/eval/token_t045/ruler8k_splits/a/lm_eval/__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: 20.0, num_fewshot: None, batch_size: 1
	\| Tasks \|Version\|Filter\|n-shot\|Metric\| \| Value \| \|Stderr\|
	\|---------------\|------:\|------\|-----:\|-----:\|---\|------:\|---\|------\|
	\|niah_multikey_2\| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 1.0000\|± \| N/A\|
	\|niah_multiquery\| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 0.9875\|± \| N/A\|
	\|niah_single_1 \| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 1.0000\|± \| N/A\|
	\|niah_single_3 \| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 0.9500\|± \| N/A\|
	\|ruler_fwe \| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 0.7667\|± \| N/A\|
	\|ruler_qa_hotpot\| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 0.4000\|± \| N/A\|
	\|ruler_vt \| 1\|none \| 0\| 4096\|↑ \|-1.0000\|± \| N/A\|
	\| \| \|none \| 0\| 8192\|↑ \| 0.8700\|± \| N/A\|