Hugging Face's logo

Spaces:
CoDynamicsLab
/
README
Running

App Files Community
Madwand1 commited on
Commit
aca2ca3
·
verified ·
1 Parent(s): fa861f6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -11
README.md CHANGED
@@ -5,8 +5,6 @@ colorFrom: blue
5
  colorTo: indigo
6
  sdk: static
7
  pinned: true
8
- thumbnail: >-
9
- https://cdn-uploads.huggingface.co/production/uploads/694646634c20c7f3d0f2eaf3/7CLHvdkIhItLCqpe7_u3W.png
10
  ---
11
 
12
  <div align="center">
@@ -35,10 +33,10 @@ The result: responses that begin in under 120 milliseconds, infrastructure costs
35
 
36
  | Model Family | Status | Avg. TTFT Speedup | E2E Speedup | Multi-Doc Pass Rate |
37
  |---|---|---|---|---|
38
- | **Qwen 2.5 14B** | ✅ Production Ready | **42.9×** | **12.9×** | **91.7% (11/12)** |
39
- | **Mistral Nemo 12B** | ✅ Verified | **104.0×** | TBD | **83.3% (10/12)** |
40
- | **Llama 3.1 8B** | ✅ Verified | **116.3×** | TBD | **83.3% (10/12)** |
41
- | **DeepSeek R1 Distill** | 🔄 In Training | *Pending* | *Pending* | *Pending* |
42
 
43
  > **Headline:** Time-To-First-Token on Qwen 2.5 14B reduced from **23.1s → 0.11s** (210× improvement on cold load).
44
  > **Persistent cache reload:** **0.0016s** — a 246× speedup over standard re-ingestion.
@@ -73,14 +71,12 @@ For teams running high-volume document analysis — M&A due diligence, legal rev
73
  **🔒 Self-Hosted License — $79**
74
  Locked model weights + inference runtime for your own A100/H100 infrastructure. Data never leaves your environment. License key delivery via [Gumroad](#) after purchase.
75
 
76
- **☁️ Managed Instance — from $5.00/hr**
77
  Spin up a LATCH-ready GPU instance directly. Includes batch JSON query interface — upload your document set, submit a structured prompt list, export results. Billed by wall-clock second. Coming soon.
78
 
79
  ---
80
 
81
- ## Architecture: CDLaC + LATCH
82
-
83
- The LATCH compilation method and neural representation format are proprietary.
84
 
85
  ---
86
 
@@ -98,4 +94,4 @@ CoDynamics Lab Corporation operates under a **Proprietary & Commercial Licensing
98
  <strong>Commercial Inquiries & Gated Access Requests</strong><br/>
99
  <a href="mailto:mike@codynamicslab.com">mike@codynamicslab.com</a> &nbsp;|&nbsp;
100
  <a href="https://www.codynamicslab.com">www.codynamicslab.com</a>
101
- </div>
 
5
  colorTo: indigo
6
  sdk: static
7
  pinned: true
 
 
8
  ---
9
 
10
  <div align="center">
 
33
 
34
  | Model Family | Status | Avg. TTFT Speedup | E2E Speedup | Multi-Doc Pass Rate |
35
  |---|---|---|---|---|
36
+ | **Qwen 2.5 14B** | ✅ Production Ready | **42.9×** | **5.2×** | **91.7% (11/12)** |
37
+ | **Mistral Nemo 12B** | ✅ Verified | **104.0×** | **19.7×** | **83.3% (10/12)** |
38
+ | **Llama 3.1 8B** | ✅ Verified | **116.3×** | **12.9×** | **83.3% (10/12)** |
39
+ | **DeepSeek R1 Distill** | ✅ Verified | **43.0×** | **3.7×** | **75.0% (9/12)** |
40
 
41
  > **Headline:** Time-To-First-Token on Qwen 2.5 14B reduced from **23.1s → 0.11s** (210× improvement on cold load).
42
  > **Persistent cache reload:** **0.0016s** — a 246× speedup over standard re-ingestion.
 
71
  **🔒 Self-Hosted License — $79**
72
  Locked model weights + inference runtime for your own A100/H100 infrastructure. Data never leaves your environment. License key delivery via [Gumroad](#) after purchase.
73
 
74
+ **☁️ Managed Instance — from $5.00 A100 / $10.00 H100 /hr**
75
  Spin up a LATCH-ready GPU instance directly. Includes batch JSON query interface — upload your document set, submit a structured prompt list, export results. Billed by wall-clock second. Coming soon.
76
 
77
  ---
78
 
79
+ The LATCH compilation method and neural representation format are proprietary to CoDynamics Lab Corporation.
 
 
80
 
81
  ---
82
 
 
94
  <strong>Commercial Inquiries & Gated Access Requests</strong><br/>
95
  <a href="mailto:mike@codynamicslab.com">mike@codynamicslab.com</a> &nbsp;|&nbsp;
96
  <a href="https://www.codynamicslab.com">www.codynamicslab.com</a>
97
+ </div>