Compare model versions
Earlier snapshot
Snapshot #264
Repository revision
b427664 (opens in a new tab)
Recorded at 04 Sep 2026, 14:07 UTC
Unchanged through 11 Sep 2026, 07:17 UTC
Later snapshot
Snapshot #333
Repository revision
12bf65d (opens in a new tab)
Recorded at 14 Sep 2026, 06:27 UTC
Summary
Licence
UnchangedNo change
CC-BY-4.0
CC-BY-4.0
PinDiff reports the declared licence and licence files. It does not interpret licence terms.
Weights
UnchangedNo change
Remote code / configuration
UnchangedNo change
Remote code
No
No
Access / gating
UnchangedNo change
Public
Public
Model card
ChangedModel card modified (2 sections)
Metadata (front matter) Changed Material section +13 −4
Diff
⋯ 17 unchanged lines
- encoder-decoder
- translation
transcribe_cpp:
Added: + schema_version: 2
Added: + wer_fleurs_de:
Added: + q8_0: 6.13
Added: + wer_fleurs_en:
Added: + q8_0: 4.75
Added: + wer_fleurs_es:
Added: + q8_0: 6.73
Added: + wer_fleurs_fr:
Added: + q8_0: 7.22
wer_librispeech_test_clean:
f32: 1.62
f16: 1.62
⋯ 2 unchanged lines
q5_k_m: 1.64
q4_k_m: 1.59
rtf_m4_max:
Removed: - metal: 99.6
Removed: - cpu: 20.4
Added: + cpu: 20.49
Added: + metal: 101.12
rtf_ryzen_4750u:
Removed: - vulkan: 14.4
Removed: - cpu: 6.9
Added: + cpu: 6.93
Added: + vulkan: 14.45
streaming: false
translate: true
lang_detect: false
Before
license: cc-by-4.0
base_model: nvidia/canary-1b-flash
base_model_relation: quantized
library_name: transcribe.cpp
pipeline_tag: automatic-speech-recognition
language:
- en
- de
- es
- fr
tags:
- gguf
- transcribe.cpp
- asr
- speech-to-text
- canary
- multitask-aed
- encoder-decoder
- translation
transcribe_cpp:
wer_librispeech_test_clean:
f32: 1.62
f16: 1.62
q8_0: 1.62
q6_k: 1.65
q5_k_m: 1.64
q4_k_m: 1.59
rtf_m4_max:
metal: 99.6
cpu: 20.4
rtf_ryzen_4750u:
vulkan: 14.4
cpu: 6.9
streaming: false
translate: true
lang_detect: false
timestamps: none
After
license: cc-by-4.0
base_model: nvidia/canary-1b-flash
base_model_relation: quantized
library_name: transcribe.cpp
pipeline_tag: automatic-speech-recognition
language:
- en
- de
- es
- fr
tags:
- gguf
- transcribe.cpp
- asr
- speech-to-text
- canary
- multitask-aed
- encoder-decoder
- translation
transcribe_cpp:
schema_version: 2
wer_fleurs_de:
q8_0: 6.13
wer_fleurs_en:
q8_0: 4.75
wer_fleurs_es:
q8_0: 6.73
wer_fleurs_fr:
q8_0: 7.22
wer_librispeech_test_clean:
f32: 1.62
f16: 1.62
q8_0: 1.62
q6_k: 1.65
q5_k_m: 1.64
q4_k_m: 1.59
rtf_m4_max:
cpu: 20.49
metal: 101.12
rtf_ryzen_4750u:
cpu: 6.93
vulkan: 14.45
streaming: false
translate: true
lang_detect: false
timestamps: none
Downloads Changed +10 −9
Diff
| Quantization | Download | Size | WER (LibriSpeech test-clean) |
| --- | --- | ---: | ---: |
Removed: - | F32 | [canary-1b-flash-F32.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F32.gguf) | 3.3 GB | 1.62% |
Removed: - | F16 | [canary-1b-flash-F16.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F16.gguf) | 1.7 GB | 1.62% |
Removed: - | Q8_0 | [canary-1b-flash-Q8_0.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q8_0.gguf) | 1.0 GB | 1.62% |
Removed: - | Q6_K | [canary-1b-flash-Q6_K.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q6_K.gguf) | 818 MB | 1.65% |
Removed: - | Q5_K_M | [canary-1b-flash-Q5_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q5_K_M.gguf) | 734 MB | 1.64% |
Removed: - | Q4_K_M | [canary-1b-flash-Q4_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q4_K_M.gguf) | 646 MB | 1.59% |
Added: + | F32 | [canary-1b-flash-F32.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F32.gguf) | 3.56 GB | 1.62% |
Added: + | F16 | [canary-1b-flash-F16.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F16.gguf) | 1.79 GB | 1.62% |
Added: + | Q8_0 | [canary-1b-flash-Q8_0.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q8_0.gguf) | 1.05 GB | 1.62% |
Added: + | Q6_K | [canary-1b-flash-Q6_K.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q6_K.gguf) | 858 MB | 1.65% |
Added: + | Q5_K_M | [canary-1b-flash-Q5_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q5_K_M.gguf) | 770 MB | 1.64% |
Added: + | Q4_K_M | [canary-1b-flash-Q4_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q4_K_M.gguf) | 677 MB | 1.59% |
Removed: - WER measured on the full LibriSpeech test-clean split (2620 utterances)
Removed: - with greedy decoding and no external LM. F32 reference baseline: 1.62%.
Removed: - NVIDIA's self-reported number on the upstream model card is 1.48%.
Added: + WER on the full LibriSpeech test-clean split (2,620 utterances), batch size 1, timestamps none. Figures without a commit were published before provenance was recorded.
Added: +
Added: + Greedy decoding, no external LM. F32 reference baseline: 1.62%. NVIDIA's
Added: + self-reported number on the upstream model card is 1.48%.
Before
| Quantization | Download | Size | WER (LibriSpeech test-clean) | | --- | --- | ---: | ---: | | F32 | [canary-1b-flash-F32.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F32.gguf) | 3.3 GB | 1.62% | | F16 | [canary-1b-flash-F16.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F16.gguf) | 1.7 GB | 1.62% | | Q8_0 | [canary-1b-flash-Q8_0.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q8_0.gguf) | 1.0 GB | 1.62% | | Q6_K | [canary-1b-flash-Q6_K.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q6_K.gguf) | 818 MB | 1.65% | | Q5_K_M | [canary-1b-flash-Q5_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q5_K_M.gguf) | 734 MB | 1.64% | | Q4_K_M | [canary-1b-flash-Q4_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q4_K_M.gguf) | 646 MB | 1.59% | WER measured on the full LibriSpeech test-clean split (2620 utterances) with greedy decoding and no external LM. F32 reference baseline: 1.62%. NVIDIA's self-reported number on the upstream model card is 1.48%.
After
| Quantization | Download | Size | WER (LibriSpeech test-clean) | | --- | --- | ---: | ---: | | F32 | [canary-1b-flash-F32.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F32.gguf) | 3.56 GB | 1.62% | | F16 | [canary-1b-flash-F16.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-F16.gguf) | 1.79 GB | 1.62% | | Q8_0 | [canary-1b-flash-Q8_0.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q8_0.gguf) | 1.05 GB | 1.62% | | Q6_K | [canary-1b-flash-Q6_K.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q6_K.gguf) | 858 MB | 1.65% | | Q5_K_M | [canary-1b-flash-Q5_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q5_K_M.gguf) | 770 MB | 1.64% | | Q4_K_M | [canary-1b-flash-Q4_K_M.gguf](https://huggingface.co/handy-computer/canary-1b-flash-gguf/resolve/main/canary-1b-flash-Q4_K_M.gguf) | 677 MB | 1.59% | WER on the full LibriSpeech test-clean split (2,620 utterances), batch size 1, timestamps none. Figures without a commit were published before provenance was recorded. Greedy decoding, no external LM. F32 reference baseline: 1.62%. NVIDIA's self-reported number on the upstream model card is 1.48%.
39 other sections unchanged.
Repository status
ChangedRepository revision moved
- Earlier
-
Revision
b427664(opens in a new tab) · 8 files · 8.70 GB - Later
-
Revision
12bf65d(opens in a new tab) · 8 files · 8.70 GB
A new revision does not by itself mean files changed; the categories above show what differs.
Files
UnchangedNo change
Evidence observed by PinDiff from the Hugging Face API and stored when recorded. PinDiff compares stored metadata, hashes and text files; it does not download weight files. Change events recorded between these snapshots