storytold/cloud-worker
2
0
Fork 0
mirror of https://github.com/storytold/cloud-worker.git synced 2026-10-09 00:09:43 +00:00
Code Issues Packages Projects Releases Wiki Activity

WEIGHTS.md: all 9 files persistent on 1TB volume, tmpfs scheme retired

Browse Source 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Hanashi
2026-08-06 02:12:42 -04:00
parent cc111c0c33
commit c3cda1fb30
1 changed files with 18 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/WEIGHTS.md
+18 -22
View File
@@ -18,36 +18,29 @@ ComfyUI finds both storage locations below via `/ComfyUI/extra_model_paths.yaml`
## Storage locations on the pod ## Storage locations on the pod
| Path | Medium | Notes | | Path | Medium | Notes |
|---------------------------------|-----------------------|---------------------------------------------------------------------------------------------------------------------| |---------------------------------|------------------------------|----------------------------------------------|
| `/workspace/minimax-h3/models/` | RunPod network volume | **Persistent.** The volume has a ~100 GB quota and is effectively full. | | `/workspace/minimax-h3/models/` | RunPod network volume (1 TB) | **Persistent** — survives pod resets/edits. |
| `/dev/shm/minimax-h3-models/` | tmpfs (RAM, 176 GB) | **EPHEMERAL** — wiped on pod restart. Refill with `bash /workspace/minimax-h3/download-weights-tmpfs.sh` (~10 min). |
## Downloaded — persistent, `/workspace/minimax-h3/models/` The volume was resized 100 GB → 1 TB on 2026-08-06 (which reset the pod), so
the earlier `/dev/shm` tmpfs overflow scheme is retired: **everything now
lives on the volume.**
## Downloaded — all files, `/workspace/minimax-h3/models/` (246 GB total)
| File | Family | Size | | File | Family | Size |
|----------------------------------------------------------------------|----------------------------------------|--------| |----------------------------------------------------------------------|----------------------------------------|--------|
| `diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors` | **int8 pruned** (ref2va) | 21 GB | | `diffusion_models/minimax_h3_fl2va_bf16.safetensors` | **bf16** (fl2va) | 66 GB |
| `diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors` | **int8 pruned** (fl2va) | 21 GB | | `diffusion_models/minimax_h3_ref2va_bf16.safetensors` | **bf16** (ref2va) | 66 GB |
| `diffusion_models/minimax_h3_fl2va_int8_convrot.safetensors` | **int8** (fl2va) | 34 GB |
| `diffusion_models/minimax_h3_ref2va_int8_convrot.safetensors` | **int8** (ref2va) | 34 GB | | `diffusion_models/minimax_h3_ref2va_int8_convrot.safetensors` | **int8** (ref2va) | 34 GB |
| `diffusion_models/minimax_h3_fl2va_pruned_int8_convrot.safetensors` | **int8 pruned** (fl2va) | 21 GB |
| `diffusion_models/minimax_h3_ref2va_pruned_int8_convrot.safetensors` | **int8 pruned** (ref2va) | 21 GB |
| `text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors` | text encoder (Qwen3-VL-32B, nvfp4 AWQ) | 16 GB | | `text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors` | text encoder (Qwen3-VL-32B, nvfp4 AWQ) | 16 GB |
| `vae/minimax_h3_video_vae_fp16.safetensors` | video VAE fp16 | 5.2 GB | | `vae/minimax_h3_video_vae_fp16.safetensors` | video VAE fp16 | 5.2 GB |
| `vae/minimax_h3_audio_vae_fp32.safetensors` | audio VAE fp32 | 0.6 GB | | `vae/minimax_h3_audio_vae_fp32.safetensors` | audio VAE fp32 | 0.6 GB |
## Downloaded — ephemeral tmpfs, `/dev/shm/minimax-h3-models/` **In progress: none.** All nine files fully downloaded and byte-verified.
| File | Family | Size |
|--------------------------------------------------------------|-------------------|-------|
| `diffusion_models/minimax_h3_fl2va_int8_convrot.safetensors` | **int8** (fl2va) | 34 GB |
| `diffusion_models/minimax_h3_ref2va_bf16.safetensors` | **bf16** (ref2va) | 66 GB |
| `diffusion_models/minimax_h3_fl2va_bf16.safetensors` | **bf16** (fl2va) | 66 GB |
The int8 family is split across the two locations only because of the volume
quota: ref2va_int8 landed on the volume before it filled; fl2va_int8 and both
bf16 files overflowed to tmpfs.
**In progress: none.** All nine files above are fully downloaded (sizes
byte-verified against the HF API).
## Not downloaded (exist upstream in the same repo) ## Not downloaded (exist upstream in the same repo)
@@ -55,7 +48,7 @@ byte-verified against the HF API).
|----------------------------------------------------------------------------|------------|--------------------------------------------------------------------------------------------| |----------------------------------------------------------------------------|------------|--------------------------------------------------------------------------------------------|
| `diffusion_models/minimax_h3_{fl2va,ref2va}_pruned_bf16.safetensors` | 40 GB each | a 4th family (pruned, unquantized) — not in the 3 requested families | | `diffusion_models/minimax_h3_{fl2va,ref2va}_pruned_bf16.safetensors` | 40 GB each | a 4th family (pruned, unquantized) — not in the 3 requested families |
| `diffusion_models/minimax_h3_{fl2va,ref2va}_pruned_fp8_scaled.safetensors` | 21 GB each | a 5th family — same size class as pruned int8 | | `diffusion_models/minimax_h3_{fl2va,ref2va}_pruned_fp8_scaled.safetensors` | 21 GB each | a 5th family — same size class as pruned int8 |
| `text_encoders/qwen3vl_32b_minimax_h3_bf16.safetensors` | 52 GB | every official template uses the nvfp4_awq TE; bf16 TE would also not fit the volume quota | | `text_encoders/qwen3vl_32b_minimax_h3_bf16.safetensors` | 52 GB | every official template uses the nvfp4_awq TE (now fits — grab if quality A/B wanted) |
| `text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors` | 27 GB | same | | `text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors` | 27 GB | same |
## History / gotchas ## History / gotchas
@@ -69,3 +62,6 @@ byte-verified against the HF API).
- After any **pod restart**: tmpfs weights are gone; ComfyUI will still boot - After any **pod restart**: tmpfs weights are gone; ComfyUI will still boot
(extra_model_paths tolerates the missing dir) but the bf16/fl2va-int8 entries (extra_model_paths tolerates the missing dir) but the bf16/fl2va-int8 entries
disappear from loader dropdowns until `download-weights-tmpfs.sh` is re-run. disappear from loader dropdowns until `download-weights-tmpfs.sh` is re-run.
- 2026-08-06 (later): volume resized to 1 TB → pod reset wiped the container
disk (ComfyUI reinstalled via `install-minimax-h3.sh`); the three overflow
files were re-downloaded to the volume. Final state: all weights persistent.