Datasets:

AINovice2005
/

raw_benchmark_results_hqq

Modalities:

Formats:

Size:

Libraries:

Dataset card Data Studio Files Files and versions

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (1)

train · 14 rows

run int64 1 1	model_id large_stringclasses 1 value	prompt_target_length int64 32 128	actual_prompt_length int64 38 134	generation_length int64 64 4.1k	prefill_latency_s float64 0.01 29.5	prefill_peak_memory_gb float64 1.05 1.32	decode_time_s float64 0.65 111	decode_tokens_per_sec float64 0.57 99.9	avg_decode_latency_per_token_ms float64 10 1.74k	peak_memory_gb float64 1.07 1.5	decode_memory_growth_gb float64 0 0.28	memory_per_generated_token_mb float64 0 0.28	hqq_backend large_stringclasses 1 value	torch_compile_backend large_stringclasses 1 value	torch_compile_mode large_stringclasses 1 value	quantization large_stringclasses 1 value	dtype large_stringclasses 1 value
1	meta-llama/Llama-3.2-1B-Instruct	32	38	64	29.528653	1.052423	111.462664	0.574183	1,741.60412	1.070037	0.017614	0.281822	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	128	0.016282	1.069691	1.289511	99.262422	10.074306	1.076385	0.006694	0.053551	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	256	0.014518	1.07372	2.563003	99.882833	10.01173	1.088547	0.014827	0.059309	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	512	0.014402	1.080844	5.141104	99.589505	10.041219	1.113739	0.032895	0.06579	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	1,024	0.01445	1.097157	10.28117	99.599559	10.040205	1.175817	0.07866	0.07866	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	2,048	0.014453	1.136926	20.594606	99.443516	10.05596	1.265023	0.128097	0.064049	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	32	38	4,096	0.014619	1.190749	41.355841	99.04284	10.096641	1.467659	0.27691	0.069228	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	64	0.016386	1.324334	0.648266	98.724844	10.129163	1.324334	0	0	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	128	0.016553	1.109348	1.293551	98.952436	10.105865	1.111925	0.002577	0.020618	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	256	0.015252	1.113254	2.581773	99.156652	10.085052	1.123696	0.010442	0.041769	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	512	0.015122	1.121067	5.155844	99.304784	10.070008	1.148256	0.027189	0.054379	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	1,024	0.015221	1.136692	10.298385	99.433069	10.057016	1.20807	0.071378	0.071378	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	2,048	0.014737	1.174371	20.592812	99.45218	10.055084	1.304444	0.130073	0.065037	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16
1	meta-llama/Llama-3.2-1B-Instruct	128	134	4,096	0.015213	1.234621	41.454579	98.806937	10.120747	1.503009	0.268388	0.067097	PYTORCH_COMPILE	inductor	reduce-overhead	HQQ_4bit	bfloat16

README.md exists but content is empty.

Downloads last month: 16

Number of rows:

Total file size: