Dataset Viewer
Auto-converted to Parquet Duplicate
task
stringclasses
1 value
model
stringclasses
6 values
params
stringclasses
4 values
quant_method
stringclasses
3 values
bits
int64
4
16
calibration_set
stringclasses
3 values
calibration_size
int64
0
512
ppl_delta_vs_fp16
float64
0
2.07
vram_gb
float64
1.9
17.4
tokens_per_sec
float64
2.97
56.3
gpu_tier
stringclasses
2 values
colab_fit
stringclasses
2 values
webgpu_runnable
stringclasses
1 value
hf_link
stringclasses
5 values
verified_date
stringdate
2026-07-30 00:00:00
2026-08-05 00:00:00
notes
stringlengths
180
562
status
stringclasses
2 values
seed
int64
0
4
calib_seq_len
int64
512
2.05k
ppl_fp16_wt2
float64
8.13
13.9
ppl_quant_wt2
float64
8.42
11
ppl_fp16_chat
float64
2.44
7.35
ppl_quant_chat
float64
2.47
4.59
ppl_ctx_len
int64
2.05k
2.05k
batch_size
int64
1
1
prompt_tokens
int64
64
64
gen_tokens
int64
256
256
artifact_gb
float64
1.04
2.08
quant_time_s
float64
0
1.53k
load_backend
stringclasses
3 values
torch_version
stringclasses
2 values
transformers_version
stringclasses
2 values
quantlib
stringclasses
2 values
error
stringclasses
16 values
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); ternar...
failed
0
512
null
null
null
null
2,048
1
64
256
null
125.5
null
2.6.0+cu124
4.51.3
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s0 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s0/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
T4
n/a
[UNVERIFIED]
null
2026-08-01
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s0 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s0/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); ternar...
failed
1
512
null
null
null
null
2,048
1
64
256
null
116.1
null
2.6.0+cu124
4.51.3
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s1 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s1/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
T4
n/a
[UNVERIFIED]
null
2026-08-01
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s1 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s1/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); ternar...
failed
2
512
null
null
null
null
2,048
1
64
256
null
114.2
null
2.6.0+cu124
4.51.3
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s2 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s2/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
awq
4
wikitext2-train
128
null
null
null
T4
n/a
[UNVERIFIED]
null
2026-08-01
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
autoawq-0.2.9
OSError("/cache/quants/bonsai-0p5b__awq__wikitext2__n128__s2 does not appear to have a file named modeling_qllama.py. Checkout 'https://huggingface.co//cache/quants/bonsai-0p5b__awq__wikitext2__n128__s2/tree/main' for available files.")
text-generation
deepgrove/Bonsai
0.5B
none-fp16
16
none
0
0
1.9
45.18
A10
n/a
[UNVERIFIED]
https://huggingface.co/deepgrove/Bonsai
2026-07-31
fp16 reference row for the awq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); ternary weights but ops run in 16-bit (NOT end-to-end ternary inference); base model is not instruction-tuned
ok
0
null
13.9169
null
7.3515
null
2,048
1
64
256
null
null
hf-fp16
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
deepgrove/Bonsai
0.5B
none-fp16
16
none
0
0
1.9
24.3
T4
yes
[UNVERIFIED]
https://huggingface.co/deepgrove/Bonsai
2026-07-31
fp16 reference row for the awq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); ternary weights but ops run in 16-bit (NOT end-to-end ternary inference); base model is not instruction-tuned
ok
0
null
13.6436
null
6.9035
null
2,048
1
64
256
null
null
hf-fp16
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
deepgrove/Bonsai
0.5B
none-fp16
16
none
0
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError("ImportError: cannot import name 'LossKwargs' from 'transformers.utils' (/usr/local/lib/python3.11/site-packages/transformers/utils/__init__.py)")
text-generation
deepgrove/Bonsai
0.5B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError("ImportError: cannot import name 'LossKwargs' from 'transformers.utils' (/usr/local/lib/python3.11/site-packages/transformers/utils/__init__.py)")
text-generation
deepgrove/Bonsai
0.5B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError("ImportError: cannot import name 'LossKwargs' from 'transformers.utils' (/usr/local/lib/python3.11/site-packages/transformers/utils/__init__.py)")
text-generation
deepgrove/Bonsai
0.5B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-07-31
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError("ImportError: cannot import name 'LossKwargs' from 'transformers.utils' (/usr/local/lib/python3.11/site-packages/transformers/utils/__init__.py)")
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a7380a5-1147618b520e6d434701c7ee;989cd5f2-aea1-490a-932a-3b2316378934)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
awq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
none-fp16
16
none
0
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737feb-1e1ac11476ccad34618f9125;d339b92d-7311-43bb-9ed6-204d39f1196d)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
none-fp16
16
none
0
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
openhermes-2.5
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737f0e-224beed13e90971b61ad5ed8;db45089a-e7c6-483f-8790-45d69fa8f334)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
128
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
32
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
0
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
1
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
meta-llama/Llama-3.2-3B-Instruct
3B
gptq
4
wikitext2-train
512
null
null
null
A10
n/a
[UNVERIFIED]
null
2026-08-05
null
failed
2
null
null
null
null
null
2,048
1
64
256
null
null
null
null
null
null
a10 batch call failed: RuntimeError('OSError: You are trying to access a gated repo.\nMake sure to have access to it at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct.\n401 Client Error. (Request ID: Root=1-6a737e44-4ebaf5c806443e120a19456d;33fd366b-2e08-4f2e-883b-b5315bafb4f7)\n\nCannot access gated repo for ...
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.9233
4.57
19.52
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
0
512
9.6394
10.5627
3.7135
3.9842
2,048
1
64
256
1.16
336.3
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.8427
4.57
8.83
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
0
512
9.0767
9.9194
3.5481
3.8013
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.8624
4.57
19.65
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
1
512
9.6394
10.5018
3.7135
3.964
2,048
1
64
256
1.16
342.7
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.7838
4.57
8.84
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
1
512
9.0767
9.8605
3.5481
3.782
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.8347
4.57
19.43
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
2
512
9.6394
10.4741
3.7135
3.9514
2,048
1
64
256
1.16
344.1
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
128
0.7538
4.57
8.97
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
2
512
9.0767
9.8305
3.5481
3.7622
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.9109
4.57
19.67
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.6394
10.5503
3.7135
3.9374
2,048
1
64
256
1.16
245.8
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.8287
4.57
8.91
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.0767
9.9054
3.5481
3.7562
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.8171
4.57
19.48
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.6394
10.4565
3.7135
3.985
2,048
1
64
256
1.16
238.7
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.7404
4.57
8.93
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.0767
9.8171
3.5481
3.8003
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.9439
4.57
19.6
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.6394
10.5833
3.7135
3.9821
2,048
1
64
256
1.16
237.9
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
32
0.8562
4.57
8.96
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=32 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.0767
9.9329
3.5481
3.7999
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.9539
4.57
19.6
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
0
512
9.6394
10.5933
3.7135
3.9495
2,048
1
64
256
1.16
838.4
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.8714
4.57
8.97
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
0
512
9.0767
9.9481
3.5481
3.7645
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.9141
4.57
19.19
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
1
512
9.6394
10.5535
3.7135
3.9227
2,048
1
64
256
1.16
793.4
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.8463
4.57
8.98
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
1
512
9.0767
9.923
3.5481
3.7413
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.8928
4.57
19.5
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
2
512
9.6394
10.5322
3.7135
3.9245
2,048
1
64
256
1.16
812.1
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
openhermes-2.5
512
0.818
4.57
8.95
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=openhermes n=512 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artif...
ok
2
512
9.0767
9.8947
3.5481
3.74
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.7852
4.57
19.74
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.6394
10.4246
3.7135
3.9545
2,048
1
64
256
1.16
359.7
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.7208
4.57
8.75
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.0767
9.7975
3.5481
3.7658
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.7789
4.57
19.38
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.6394
10.4183
3.7135
3.9818
2,048
1
64
256
1.16
341.1
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.7106
4.57
8.75
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.0767
9.7873
3.5481
3.7958
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.8541
4.57
19.59
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.6394
10.4935
3.7135
3.9114
2,048
1
64
256
1.16
345.8
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
128
0.7722
4.57
8.74
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.0767
9.8489
3.5481
3.7216
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.7396
4.57
19.58
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
0
512
9.6394
10.379
3.7135
3.9294
2,048
1
64
256
1.16
251
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.6713
4.57
8.7
T4
yes
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
0
512
9.0767
9.748
3.5481
3.7407
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.7806
4.57
19.32
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
1
512
9.6394
10.42
3.7135
3.9339
2,048
1
64
256
1.16
241.9
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.7046
4.57
8.74
T4
yes
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
1
512
9.0767
9.7813
3.5481
3.7472
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.7727
4.57
19.18
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
2
512
9.6394
10.4121
3.7135
3.9857
2,048
1
64
256
1.16
248.7
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
32
0.6887
4.57
8.75
T4
yes
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=32 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifac...
ok
2
512
9.0767
9.7654
3.5481
3.7983
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.7489
4.57
19.72
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.6394
10.3883
3.7135
3.93
2,048
1
64
256
1.16
831.8
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.6671
4.57
8.71
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
0
512
9.0767
9.7438
3.5481
3.7481
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.7514
4.57
19.55
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.6394
10.3908
3.7135
3.9093
2,048
1
64
256
1.16
770.4
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.6705
4.57
8.71
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
1
512
9.0767
9.7472
3.5481
3.7295
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.7269
4.57
19.44
A10
n/a
[UNVERIFIED]
null
2026-07-30
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.6394
10.3663
3.7135
3.9235
2,048
1
64
256
1.16
777.9
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
awq
4
wikitext2-train
512
0.6469
4.57
8.71
T4
yes
[UNVERIFIED]
null
2026-07-31
AWQ zero_point g128 GEMM calib_len=512; autoawq 0.2.9 (upstream-deprecated; pinned to its last-tested torch2.6.0/transformers4.51.3 combo); calib=wikitext2 n=512 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifa...
ok
2
512
9.0767
9.7236
3.5481
3.7385
2,048
1
64
256
1.16
null
gemm
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
none-fp16
16
none
0
0
6.48
45.38
A10
n/a
[UNVERIFIED]
https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct
2026-07-30
fp16 reference row for the awq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill)
ok
0
null
9.6394
null
3.7135
null
2,048
1
64
256
null
null
hf-fp16
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
none-fp16
16
none
0
0
6.48
32.64
T4
yes
[UNVERIFIED]
https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct
2026-07-30
fp16 reference row for the awq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill)
ok
0
null
9.0767
null
3.5481
null
2,048
1
64
256
null
null
hf-fp16
2.6.0+cu124
4.51.3
autoawq-0.2.9
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
none-fp16
16
none
0
0
6.48
35.52
A10
n/a
[UNVERIFIED]
https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct
2026-07-30
fp16 reference row for the gptq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill)
ok
0
null
9.6394
null
3.7135
null
2,048
1
64
256
null
null
hf-fp16
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
none-fp16
16
none
0
0
6.48
28.3
T4
yes
[UNVERIFIED]
https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct
2026-07-30
fp16 reference row for the gptq stack; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill)
ok
0
null
9.0766
null
3.5483
null
2,048
1
64
256
null
null
hf-fp16
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.9359
7.69
20.3
A10
n/a
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s0 (not pushed to H...
ok
0
2,048
9.6394
10.5753
3.7135
3.874
2,048
1
64
256
1.16
344.9
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.8516
7.92
3.88
T4
yes
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s0 (not pushed to H...
ok
0
2,048
9.0766
9.9282
3.5483
3.6796
2,048
1
64
256
1.16
null
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.8609
7.69
20.73
A10
n/a
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s1 (not pushed to H...
ok
1
2,048
9.6394
10.5003
3.7135
3.8874
2,048
1
64
256
1.16
347.9
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.8042
7.92
3.9
T4
yes
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s1 (not pushed to H...
ok
1
2,048
9.0766
9.8808
3.5483
3.7019
2,048
1
64
256
1.16
null
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.8055
7.68
20.83
A10
n/a
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s2 (not pushed to H...
ok
2
2,048
9.6394
10.4449
3.7135
3.8637
2,048
1
64
256
1.16
356
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
128
0.7546
7.92
3.9
T4
yes
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=128 seed=2; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n128__s2 (not pushed to H...
ok
2
2,048
9.0766
9.8312
3.5483
3.6747
2,048
1
64
256
1.16
null
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
32
1.0467
6.05
21.66
A10
n/a
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n32__s0 (not pushed to HF ...
ok
0
2,048
9.6394
10.6861
3.7135
3.9038
2,048
1
64
256
1.16
255.6
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
32
0.9662
5.76
4
T4
yes
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=32 seed=0; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n32__s0 (not pushed to HF ...
ok
0
2,048
9.0766
10.0428
3.5483
3.7166
2,048
1
64
256
1.16
null
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
32
0.9647
7.13
21.16
A10
n/a
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n32__s1 (not pushed to HF ...
ok
1
2,048
9.6394
10.6041
3.7135
3.8804
2,048
1
64
256
1.16
230.1
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
text-generation
Qwen/Qwen2.5-1.5B-Instruct
1.5B
gptq
4
openhermes-2.5
32
0.8918
6.84
3.93
T4
yes
[UNVERIFIED]
null
2026-07-30
GPTQ g128 desc_act=True damp=0.01 calib_len=2048; calib=openhermes n=32 seed=1; ppl: wikitext2-test + OpenHermes-heldout, ctx2048 non-overlapping, token-capped per tier; tok/s: greedy b1 64->256 median-of-3 (incl. prefill); artifact modal vol quant-bench:/quants/qwen25-1p5b__gptq__openhermes__n32__s1 (not pushed to HF ...
ok
1
2,048
9.0766
9.9684
3.5483
3.6887
2,048
1
64
256
1.16
null
torch-fallback
2.8.0+cu128
5.14.1
gptqmodel-7.3.2
null
End of preview. Expand in Data Studio

QuantBench leaderboard data

Raw benchmark data behind the QuantBench leaderboard: calibration-quality GPTQ/AWQ quantization results across model sizes, calibration corpora, and GPU tiers. 331 rows (239 ok / 92 failed — failed runs are published too; a documented failure is a finding, not noise).

Models

  • Qwen/Qwen2.5-1.5B-Instruct (1.5B)
  • HuggingFaceTB/SmolLM2-1.7B-Instruct (1.7B)
  • deepgrove/Bonsai (0.5B)
  • Qwen/Qwen2.5-3B-Instruct (3B) — licence pending, rows only, no weights here
  • HuggingFaceTB/SmolLM3-3B (3B)

Files

  • rows.csv — the full table (see the leaderboard site for column definitions and the methodology page for the eval protocol).
  • rows.parquet — the same table with an explicit Arrow schema (numeric columns nullable int64/double, blanks as null; everything else string). This is what the dataset viewer reads (configs: above); rows.csv is unchanged for humans.
  • <row_id>.json — one file per row, same fields as its rows.csv line.
  • niche_rows.csv — task-level niche columns over the same artifacts (HumanEval[0:50] pass@1, FiQA-2018 sentiment accuracy [flare-finqa is gated for this token — instrument recorded per row], SEC-filing context drift, 10-prompt ROUGE-L consistency), each vs a same-stack fp16 baseline; failures/skips recorded as rows.
  • logs/ — the driver log, last heartbeat, and spend ledger from the run that produced this data (transparency, not polish — these are raw operational logs).
  • pools/BUILD.md — exact recipe to reproduce the calibration/eval text pools. No third-party corpus text is redistributed here.

Honesty notes carried into every row

  • GPTQ rows in this sweep ran gptqmodel's torch-fallback kernel path (load_backend column), not the optimized kernel — perplexity is unaffected, throughput numbers understate what a working kernel would show.
  • fp16 baselines are per-stack (autoawq vs gptqmodel use different torch/transformers pins), so ppl_delta_vs_fp16 is only comparable within one stack and one GPU tier, never across them.
  • webgpu_runnable is [UNVERIFIED] on every row — no browser measurement was run.
  • Machine-generated benchmark output. Independent verification welcome; treat any single row as a data point, not a certified result.
Downloads last month
819