FINAL-Bench/POCKET-Image-Zimage
Text-to-Image β’ 3B β’ Updated β’ 110 β’ 33
Minseo Kim
MinseoKim-03
AI & ML interests
None yet
Recent Activity
liked a model 5 days ago
FINAL-Bench/POCKET-Image-Zimage liked a Space 5 days ago
FINAL-Bench/POCKET-Image-Studio reacted to SeaWolf-AI's post with π 5 days ago
πΌοΈ POCKET-Image β the POCKET series goes visual: character-perfect text in any language, on-device
A new model in VIDRAFT's POCKET family. POCKET put 35B-class models on phones and no-GPU PCs. POCKET-Image carries the same "big capability, small hardware" idea into image generation β and fixes the one thing nearly every image model gets wrong: text.
Type "μλ
νμΈμ" into a typical model and you get "μγ
κΈ°." Hangul alone composes 11,172 syllable blocks; Arabic connects its letters; Thai stacks marks. Diffusion models draw scripts as shapes, so they smear. POCKET-Image renders every glyph exactly β νκ΅μ΄ Β· δΈζ Β· ζ₯ζ¬θͺ Β· Ψ§ΩΨΉΨ±Ψ¨ΩΨ© (RTL) Β· ΰΉΰΈΰΈ’ Β· Latin and more β onto any scene you describe.
What it is:
β’ 100% accurate text, any language β where global models produce gibberish
β’ Any background from a prompt β text is optional (empty β a pure image)
β’ No GPU, no NPU β runs on plain CPU + RAM via the POCKET-Core engine
β’ Measured footprint: 8.6 GB (RTX 3050/4060) Β· 4.5 GB (offloaded, 6 GB cards) Β· 13.4 GB (MacBook, 16 GB+)
β’ Windows Β· macOS Β· Linux Β· fully local, no cloud
Built on the open, commercial-friendly Z-Image (Apache-2.0) foundation.
Honest note: the text is the guaranteed-correct part β the surrounding scene is ordinary generation, so a busy foreground can crowd the letters. We say so; clean backgrounds stay razor-sharp.
π¨ Studio β generate right here, any language:
https://huggingface.co/spaces/FINAL-Bench/POCKET-Image-Studio
𧩠Model card:
https://huggingface.co/FINAL-Bench/POCKET-Image-Zimage
π The POCKET collection:
https://huggingface.co/collections/FINAL-Bench/pocket-modelsOrganizations
None yet
liked a
model 5 days ago
liked a
Space 5 days ago
reacted to SeaWolf-AI's post with π 5 days ago
Post
6363
πΌοΈ POCKET-Image β the POCKET series goes visual: character-perfect text in any language, on-device
A new model in VIDRAFT's POCKET family. POCKET put 35B-class models on phones and no-GPU PCs. POCKET-Image carries the same "big capability, small hardware" idea into image generation β and fixes the one thing nearly every image model gets wrong: text.
Type "μλ νμΈμ" into a typical model and you get "μγ κΈ°." Hangul alone composes 11,172 syllable blocks; Arabic connects its letters; Thai stacks marks. Diffusion models draw scripts as shapes, so they smear. POCKET-Image renders every glyph exactly β νκ΅μ΄ Β· δΈζ Β· ζ₯ζ¬θͺ Β· Ψ§ΩΨΉΨ±Ψ¨ΩΨ© (RTL) Β· ΰΉΰΈΰΈ’ Β· Latin and more β onto any scene you describe.
What it is:
β’ 100% accurate text, any language β where global models produce gibberish
β’ Any background from a prompt β text is optional (empty β a pure image)
β’ No GPU, no NPU β runs on plain CPU + RAM via the POCKET-Core engine
β’ Measured footprint: 8.6 GB (RTX 3050/4060) Β· 4.5 GB (offloaded, 6 GB cards) Β· 13.4 GB (MacBook, 16 GB+)
β’ Windows Β· macOS Β· Linux Β· fully local, no cloud
Built on the open, commercial-friendly Z-Image (Apache-2.0) foundation.
Honest note: the text is the guaranteed-correct part β the surrounding scene is ordinary generation, so a busy foreground can crowd the letters. We say so; clean backgrounds stay razor-sharp.
π¨ Studio β generate right here, any language:
FINAL-Bench/POCKET-Image-Studio
𧩠Model card:
FINAL-Bench/POCKET-Image-Zimage
π The POCKET collection:
https://huggingface.co/collections/FINAL-Bench/pocket-models
A new model in VIDRAFT's POCKET family. POCKET put 35B-class models on phones and no-GPU PCs. POCKET-Image carries the same "big capability, small hardware" idea into image generation β and fixes the one thing nearly every image model gets wrong: text.
Type "μλ νμΈμ" into a typical model and you get "μγ κΈ°." Hangul alone composes 11,172 syllable blocks; Arabic connects its letters; Thai stacks marks. Diffusion models draw scripts as shapes, so they smear. POCKET-Image renders every glyph exactly β νκ΅μ΄ Β· δΈζ Β· ζ₯ζ¬θͺ Β· Ψ§ΩΨΉΨ±Ψ¨ΩΨ© (RTL) Β· ΰΉΰΈΰΈ’ Β· Latin and more β onto any scene you describe.
What it is:
β’ 100% accurate text, any language β where global models produce gibberish
β’ Any background from a prompt β text is optional (empty β a pure image)
β’ No GPU, no NPU β runs on plain CPU + RAM via the POCKET-Core engine
β’ Measured footprint: 8.6 GB (RTX 3050/4060) Β· 4.5 GB (offloaded, 6 GB cards) Β· 13.4 GB (MacBook, 16 GB+)
β’ Windows Β· macOS Β· Linux Β· fully local, no cloud
Built on the open, commercial-friendly Z-Image (Apache-2.0) foundation.
Honest note: the text is the guaranteed-correct part β the surrounding scene is ordinary generation, so a busy foreground can crowd the letters. We say so; clean backgrounds stay razor-sharp.
π¨ Studio β generate right here, any language:
FINAL-Bench/POCKET-Image-Studio
𧩠Model card:
FINAL-Bench/POCKET-Image-Zimage
π The POCKET collection:
https://huggingface.co/collections/FINAL-Bench/pocket-models
reacted to SeaWolf-AI's post with π 8 days ago
Post
3122
POCKET now speaks Gemma 4 β a 26B model that loads in every app, and runs on your PC with no GPU
We're adding a Gemma-4 sibling to POCKET: POCKET-26B, built from Google's Gemma-4-26B-A4B (Apache-2.0). Our flagship POCKET-35B is a Qwen-family MoE and needs a recent llama.cpp; POCKET-26B trades a little size for the thing people kept asking for β it just loads, everywhere, today: Ollama, LM Studio, PocketPal, MLX, any stock llama.cpp. No fork, no bleeding-edge runtime, no CUDA, no cloud.
It's a sparse Mixture-of-Experts (25.2B total, ~4B active per token), so the work per token stays small β a real 26B that generates on a CPU with no graphics card.
Two things make it stand out:
1) Universal compatibility. Gemma 4 is a standard, widely-supported architecture, so POCKET-26B runs on the tools you already have β no waiting for your app to add a new model type.
2) Quality that survives compression. Measured GPQA-Diamond (198 q, greedy):
β’ Full base: 67.7%
β’ POCKET-26B Q4_K_M (17 GB): 67.7% β lossless
β’ POCKET-26B Q2_K (11 GB): 67.2% β near-lossless, at 11 GB
Live, on a CPU-only box (our demo Space β POCKET-26B vs Bonsai-27B, same machine, same stock llama.cpp): POCKET-26B β 19 tok/s vs Bonsai β 6 tok/s β about 3Γ faster generation, no GPU. (Honest notes: shared CPU box, sequential race; a dedicated machine is faster.)
Where it fits in the family:
β’ POCKET-35B (Qwen MoE) β bigger, top-tier, needs a recent llama.cpp.
β’ POCKET-26B (Gemma 4) β loads in any app, quality-robust when compressed. The demo runs the Q4_K_M build; Q2_K (11 GB) is the smallest footprint. For a true β€8 GB phone, the 5 GB POCKET-KR (Qwen) is still the pick.
Try it and grab it:
π₯οΈ Live demo (Gemma4-based, answering on a CPU, no GPU): FINAL-Bench/POCKET-26B-CPU
π¦ POCKET-26B-GGUF (Q4_K_M 17 GB Β· Q2_K 11 GB): FINAL-Bench/POCKET-26B-GGUF
π POCKET collection: https://huggingface.co/collections/FINAL-Bench/pocket-models
We're adding a Gemma-4 sibling to POCKET: POCKET-26B, built from Google's Gemma-4-26B-A4B (Apache-2.0). Our flagship POCKET-35B is a Qwen-family MoE and needs a recent llama.cpp; POCKET-26B trades a little size for the thing people kept asking for β it just loads, everywhere, today: Ollama, LM Studio, PocketPal, MLX, any stock llama.cpp. No fork, no bleeding-edge runtime, no CUDA, no cloud.
It's a sparse Mixture-of-Experts (25.2B total, ~4B active per token), so the work per token stays small β a real 26B that generates on a CPU with no graphics card.
Two things make it stand out:
1) Universal compatibility. Gemma 4 is a standard, widely-supported architecture, so POCKET-26B runs on the tools you already have β no waiting for your app to add a new model type.
2) Quality that survives compression. Measured GPQA-Diamond (198 q, greedy):
β’ Full base: 67.7%
β’ POCKET-26B Q4_K_M (17 GB): 67.7% β lossless
β’ POCKET-26B Q2_K (11 GB): 67.2% β near-lossless, at 11 GB
Live, on a CPU-only box (our demo Space β POCKET-26B vs Bonsai-27B, same machine, same stock llama.cpp): POCKET-26B β 19 tok/s vs Bonsai β 6 tok/s β about 3Γ faster generation, no GPU. (Honest notes: shared CPU box, sequential race; a dedicated machine is faster.)
Where it fits in the family:
β’ POCKET-35B (Qwen MoE) β bigger, top-tier, needs a recent llama.cpp.
β’ POCKET-26B (Gemma 4) β loads in any app, quality-robust when compressed. The demo runs the Q4_K_M build; Q2_K (11 GB) is the smallest footprint. For a true β€8 GB phone, the 5 GB POCKET-KR (Qwen) is still the pick.
Try it and grab it:
π₯οΈ Live demo (Gemma4-based, answering on a CPU, no GPU): FINAL-Bench/POCKET-26B-CPU
π¦ POCKET-26B-GGUF (Q4_K_M 17 GB Β· Q2_K 11 GB): FINAL-Bench/POCKET-26B-GGUF
π POCKET collection: https://huggingface.co/collections/FINAL-Bench/pocket-models
- 3 replies
liked a
Space 8 days ago
liked a
model 8 days ago
liked a
Space 12 days ago
liked 4
models 12 days ago
FINAL-Bench/POCKET-EN-GGUF
Text Generation β’ 23B β’ Updated β’ 3.08k β’ 34
FINAL-Bench/POCKET-KR-MLX
Text Generation β’ 15B β’ Updated β’ 710 β’ 27
FINAL-Bench/POCKET-KR-GGUF
Text Generation β’ 23B β’ Updated β’ 1.75k β’ 26
FINAL-Bench/POCKET-35B-GGUF
Text Generation β’ 35B β’ Updated β’ 86.1k β’ 63
reacted to SeaWolf-AI's post with π 12 days ago
Post
4033
π± POCKET β a 35-billion-parameter model that runs on your iPhone, and on your PC with no GPU
We're releasing POCKET, VIDRAFT's flagship Darwin-36B-Opus compressed for on-device use. No fork, no CUDA, no cloud β it runs on stock llama.cpp. It's a sparse Mixture-of-Experts model (256 experts, only 8 active per token), so the file can be large while the work per token stays small. That's what lets a 35B model run on a phone, and generate fast on a CPU with no graphics card.
Measured (POCKET-35B IQ1_M vs Bonsai-27B Q1_0):
β’ CPU generate (Xeon, 16 threads): 27.0 vs 10.1 tok/s β 2.69Γ faster
β’ GPU generate (H100): 197 vs 89 tok/s β 2.22Γ faster
β’ GPU prompt processing (H100): 753 vs 1816 β 0.41Γ (Bonsai wins this one β MoE prefill wakes every expert, so sparsity stops helping there. We say so.)
β’ Quality (HellaSwag, 400 q): 61.0% vs 60.0% β a tie (confidence intervals overlap)
On a real consumer laptop β MacBook M3 Pro (18 GB) β POCKET wins every axis, prompt processing included:
β’ Metal generate: 25.4 vs 12.8 β 1.99Γ
β’ CPU generate: 13.8 vs 4.4 β 3.13Γ
β’ Metal prompt: 240.7 vs 73.4 β 3.28Γ
One more quiet fact: the same-size, quality-oriented rival Ternary-Bonsai-27B (7.2 GB) fails to load in upstream llama.cpp at all β it needs the PrismML fork. POCKET runs on the tools you already have: LM Studio, Ollama, PocketPal, MLX.
π Full story (tech, measurements, recipes): https://huggingface.co/blog/FINAL-Bench/pocket
Models:
π¦ POCKET-35B-GGUF (PC / server, no GPU): FINAL-Bench/POCKET-35B-GGUF
π°π· POCKET-KR-GGUF (Android): FINAL-Bench/POCKET-KR-GGUF
π POCKET-KR-MLX (iPhone / Mac): FINAL-Bench/POCKET-KR-MLX
π POCKET-EN-GGUF (English phone / PC): FINAL-Bench/POCKET-EN-GGUF
π₯οΈ Live demo (answering on a CPU, no GPU): FINAL-Bench/POCKET-35B-CPU
π Collection: FINAL-Bench/pocket-models-6a618ee5d23eafb7e185a5c6
We're releasing POCKET, VIDRAFT's flagship Darwin-36B-Opus compressed for on-device use. No fork, no CUDA, no cloud β it runs on stock llama.cpp. It's a sparse Mixture-of-Experts model (256 experts, only 8 active per token), so the file can be large while the work per token stays small. That's what lets a 35B model run on a phone, and generate fast on a CPU with no graphics card.
Measured (POCKET-35B IQ1_M vs Bonsai-27B Q1_0):
β’ CPU generate (Xeon, 16 threads): 27.0 vs 10.1 tok/s β 2.69Γ faster
β’ GPU generate (H100): 197 vs 89 tok/s β 2.22Γ faster
β’ GPU prompt processing (H100): 753 vs 1816 β 0.41Γ (Bonsai wins this one β MoE prefill wakes every expert, so sparsity stops helping there. We say so.)
β’ Quality (HellaSwag, 400 q): 61.0% vs 60.0% β a tie (confidence intervals overlap)
On a real consumer laptop β MacBook M3 Pro (18 GB) β POCKET wins every axis, prompt processing included:
β’ Metal generate: 25.4 vs 12.8 β 1.99Γ
β’ CPU generate: 13.8 vs 4.4 β 3.13Γ
β’ Metal prompt: 240.7 vs 73.4 β 3.28Γ
One more quiet fact: the same-size, quality-oriented rival Ternary-Bonsai-27B (7.2 GB) fails to load in upstream llama.cpp at all β it needs the PrismML fork. POCKET runs on the tools you already have: LM Studio, Ollama, PocketPal, MLX.
π Full story (tech, measurements, recipes): https://huggingface.co/blog/FINAL-Bench/pocket
Models:
π¦ POCKET-35B-GGUF (PC / server, no GPU): FINAL-Bench/POCKET-35B-GGUF
π°π· POCKET-KR-GGUF (Android): FINAL-Bench/POCKET-KR-GGUF
π POCKET-KR-MLX (iPhone / Mac): FINAL-Bench/POCKET-KR-MLX
π POCKET-EN-GGUF (English phone / PC): FINAL-Bench/POCKET-EN-GGUF
π₯οΈ Live demo (answering on a CPU, no GPU): FINAL-Bench/POCKET-35B-CPU
π Collection: FINAL-Bench/pocket-models-6a618ee5d23eafb7e185a5c6
- 5 replies
reacted to SeaWolf-AI's post with π₯ 14 days ago
Post
5146
A small gift for anyone building or studying foundation models.
Most "open" models hand you the weights and stop there. With Aether-7B-5Attn we wanted to hand over the whole thing β so you can actually learn from it, reproduce it, and build on it: the data recipe, the training code, every hyperparameter, the complete logs, and the intermediate checkpoints. All Apache-2.0, reproducible byte-for-byte.
What you can do with it:
π Rebuild it from scratch, or fork the recipe for your own model
π¬ Study a real heterogeneous-attention MoE β 49 layers place 5 attention mechanisms on a 7Γ7 Latin square, arranged as a clean, attributable ablation
π Trace training dynamics across the released checkpoints (110k / 115k / 162k)
It's a modest 6.59B model, and an honest one β the limitations (no KV-cache in this build, small scale) are written right in the card. We're not claiming it's special. If any piece of it saves you time or teaches you something, that's exactly what we hoped for. π€
π Full write-up β
[blog] Β· https://huggingface.co/blog/FINAL-Bench/opensource-llm
π¦ 5 Attention Base Β· FINAL-Bench/Aether-7B-5Attn
π― 5 Attention Instruct Β· FINAL-Bench/Aether-7B-5Attn-it
π 5 Attention Live demo Β· FINAL-Bench/Aether-Sovereign-AI
π¦ 7 Attention Base Β· https://huggingface.co/FINAL-Bench/Aether-7B-7Attn-base
π¦ 11 Attention Base Β· FINAL-Bench/Aether-6B-11Attn-base
𧬠Collection · https://huggingface.co/collections/FINAL-Bench/aether-foundation-model
#opensource #LLM #MoE #reproducibility #Apache2
Most "open" models hand you the weights and stop there. With Aether-7B-5Attn we wanted to hand over the whole thing β so you can actually learn from it, reproduce it, and build on it: the data recipe, the training code, every hyperparameter, the complete logs, and the intermediate checkpoints. All Apache-2.0, reproducible byte-for-byte.
What you can do with it:
π Rebuild it from scratch, or fork the recipe for your own model
π¬ Study a real heterogeneous-attention MoE β 49 layers place 5 attention mechanisms on a 7Γ7 Latin square, arranged as a clean, attributable ablation
π Trace training dynamics across the released checkpoints (110k / 115k / 162k)
It's a modest 6.59B model, and an honest one β the limitations (no KV-cache in this build, small scale) are written right in the card. We're not claiming it's special. If any piece of it saves you time or teaches you something, that's exactly what we hoped for. π€
π Full write-up β
[blog] Β· https://huggingface.co/blog/FINAL-Bench/opensource-llm
π¦ 5 Attention Base Β· FINAL-Bench/Aether-7B-5Attn
π― 5 Attention Instruct Β· FINAL-Bench/Aether-7B-5Attn-it
π 5 Attention Live demo Β· FINAL-Bench/Aether-Sovereign-AI
π¦ 7 Attention Base Β· https://huggingface.co/FINAL-Bench/Aether-7B-7Attn-base
π¦ 11 Attention Base Β· FINAL-Bench/Aether-6B-11Attn-base
𧬠Collection · https://huggingface.co/collections/FINAL-Bench/aether-foundation-model
#opensource #LLM #MoE #reproducibility #Apache2
- 5 replies
liked a
dataset 14 days ago
liked 2
models 14 days ago
liked a
Space 14 days ago
upvoted a collection 14 days ago
upvoted an article 14 days ago
Article
Aether-7B-5Attn: A 100% Open-Source Sovereign Foundation Model β and a Controlled Experiment in Heterogeneous Attention
FINAL-Bench
β’ β’ 21 reacted to SeaWolf-AI's post with π₯ 21 days ago
Post
5405
π΅ VKUE β No GPU? Runs anyway.
"Frontier models need a datacenter GPU" rests on a hidden assumption: that the model reads ALL its parameters every token. Decode is memory-bandwidth bound β sweep 34B params/token and an 8 GB card dies at 1β2 tok/s.
So we ran ONE 34.7B reasoning model β Ourbox-35B-JGOS, a sparse Mixture-of-Experts β as the identical weights across the whole hardware spectrum. All measured:
β’ B200: 18,057 tok/s (aggregate)
β’ 1Γ A10G: 126 tok/s
β’ 8 GB laptop (RTX 5060): 20 tok/s
β’ GPU-less CPU: 17 tok/s
Why it works: Ourbox holds 34.7B params but only ~3B are active per token (256 experts, top-8). Since decode is bandwidth-bound, a dense 34B moves ~16.7 GB/token while Ourbox moves ~1.45 GB β ~11Γ less traffic. Put the experts in system RAM, keep attention/router/shared on the GPU, and a 34.7B reasoner runs on an 8 GB laptop β or no GPU at all.
Sparsity alone, proven (same laptop, same quant, ~same footprint): Ourbox-35B (A3B) 20.01 tok/s vs Qwen2.5-32B (dense) 5.36 β 3.7Γ from sparsity alone, ~2Γ the best dense-32B on any 8 GB machine. Not a toy: GPQA Diamond 86.4% (maj@8).
Try it live (same prompt, GPU vs GPU-less CPU, live tok/s). Honest scope: one machine's measurements; the CPU path proves it RUNS without a GPU, not that it beats one.
π Article: https://huggingface.co/blog/FINAL-Bench/vkue
π΅ GPU vs CPU demo: https://final-bench-ourbox-35b-vkue-demo.hf.space/
π΅ CPU-only demo: https://final-bench-ourbox-35b-vkue-cpu.hf.space
π VKUE leaderboard: FINAL-Bench/VKUE
π€ Model: FINAL-Bench/Ourbox-35B-JGOS-GGUF
β‘ VKAE (speed): VIDraft/vkae
VKUE is the "runs anywhere" side of our serving line; VKAE the "fast on datacenter GPUs" side. VKAE is fast; VKUE is everywhere.
"Frontier models need a datacenter GPU" rests on a hidden assumption: that the model reads ALL its parameters every token. Decode is memory-bandwidth bound β sweep 34B params/token and an 8 GB card dies at 1β2 tok/s.
So we ran ONE 34.7B reasoning model β Ourbox-35B-JGOS, a sparse Mixture-of-Experts β as the identical weights across the whole hardware spectrum. All measured:
β’ B200: 18,057 tok/s (aggregate)
β’ 1Γ A10G: 126 tok/s
β’ 8 GB laptop (RTX 5060): 20 tok/s
β’ GPU-less CPU: 17 tok/s
Why it works: Ourbox holds 34.7B params but only ~3B are active per token (256 experts, top-8). Since decode is bandwidth-bound, a dense 34B moves ~16.7 GB/token while Ourbox moves ~1.45 GB β ~11Γ less traffic. Put the experts in system RAM, keep attention/router/shared on the GPU, and a 34.7B reasoner runs on an 8 GB laptop β or no GPU at all.
Sparsity alone, proven (same laptop, same quant, ~same footprint): Ourbox-35B (A3B) 20.01 tok/s vs Qwen2.5-32B (dense) 5.36 β 3.7Γ from sparsity alone, ~2Γ the best dense-32B on any 8 GB machine. Not a toy: GPQA Diamond 86.4% (maj@8).
Try it live (same prompt, GPU vs GPU-less CPU, live tok/s). Honest scope: one machine's measurements; the CPU path proves it RUNS without a GPU, not that it beats one.
π Article: https://huggingface.co/blog/FINAL-Bench/vkue
π΅ GPU vs CPU demo: https://final-bench-ourbox-35b-vkue-demo.hf.space/
π΅ CPU-only demo: https://final-bench-ourbox-35b-vkue-cpu.hf.space
π VKUE leaderboard: FINAL-Bench/VKUE
π€ Model: FINAL-Bench/Ourbox-35B-JGOS-GGUF
β‘ VKAE (speed): VIDraft/vkae
VKUE is the "runs anywhere" side of our serving line; VKAE the "fast on datacenter GPUs" side. VKAE is fast; VKUE is everywhere.
- 7 replies