tg-techie-agents commited on
Commit
05edd54
·
verified ·
1 Parent(s): d9c106f

latest.json: checkpoints/4522274k (step 119,408)

Browse files
Files changed (4) hide show
  1. README.md +4 -4
  2. latest.json +7 -7
  3. latest/config.json +2 -2
  4. latest/model.safetensors +1 -1
README.md CHANGED
@@ -30,7 +30,7 @@ deliberately derived from it: Nirca began as a port of mini-AGI to MLX.
30
  ➡️ **Try it in your browser (WebGPU):** https://tg-techie-agents.github.io/nirca-mini-webgpu/mini/
31
 
32
  This card is written by program from the checkpoint's own record, at step
33
- 119,196 (4,516,803,395 tokens seen).
34
 
35
  ## Model details
36
 
@@ -115,9 +115,9 @@ Each step is listed because the run's records show it was done:
115
 
116
  ## Training procedure
117
 
118
- - **Steps:** 119,196
119
- - **Tokens seen:** 4,516,803,395 (one token is one byte)
120
- - **Quantization-aware training:** from step 14,655 of 119,196, so for 99% of its 4,516,803,395 tokens
121
  - **Objective:** next-byte prediction; on chats, loss on the assistant's turns only
122
  - **Recurrence:** the number of passes is drawn per step during training, and
123
  the halting head learns when to stop
 
30
  ➡️ **Try it in your browser (WebGPU):** https://tg-techie-agents.github.io/nirca-mini-webgpu/mini/
31
 
32
  This card is written by program from the checkpoint's own record, at step
33
+ 119,408 (4,522,274,480 tokens seen).
34
 
35
  ## Model details
36
 
 
115
 
116
  ## Training procedure
117
 
118
+ - **Steps:** 119,408
119
+ - **Tokens seen:** 4,522,274,480 (one token is one byte)
120
+ - **Quantization-aware training:** from step 14,655 of 119,408, so for 99% of its 4,522,274,480 tokens
121
  - **Objective:** next-byte prediction; on chats, loss on the assistant's turns only
122
  - **Recurrence:** the number of passes is drawn per step during training, and
123
  the halting head learns when to stop
latest.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "name": "nirca-mini",
3
- "step": 119196,
4
  "quant": "ternary",
5
  "quant_scope": "all",
6
  "n_experts": 32,
@@ -40,17 +40,17 @@
40
  "pool_depth": 1,
41
  "pool_max": 32
42
  },
43
- "tokens": 4516803395,
44
- "checkpoint": "checkpoints/4516803k/",
45
  "files": {
46
  "config.json": {
47
- "url": "https://huggingface.co/nirca/nirca-mini/resolve/main/checkpoints/4516803k/config.json",
48
- "sha256": "3ce56c3e4ec3ee592e2e80d92670274ff3e61fffea8b3e9044e0edc7f3bd82df",
49
  "bytes": 874
50
  },
51
  "model.safetensors": {
52
- "url": "https://huggingface.co/nirca/nirca-mini/resolve/main/checkpoints/4516803k/model.safetensors",
53
- "sha256": "d4b3bcc71ce2e9dd7442957e7c5304b8eb85aacbb64b183194e465add039e5d8",
54
  "bytes": 32638048
55
  }
56
  }
 
1
  {
2
  "name": "nirca-mini",
3
+ "step": 119408,
4
  "quant": "ternary",
5
  "quant_scope": "all",
6
  "n_experts": 32,
 
40
  "pool_depth": 1,
41
  "pool_max": 32
42
  },
43
+ "tokens": 4522274480,
44
+ "checkpoint": "checkpoints/4522274k/",
45
  "files": {
46
  "config.json": {
47
+ "url": "https://huggingface.co/nirca/nirca-mini/resolve/main/checkpoints/4522274k/config.json",
48
+ "sha256": "a4090fd764a1afadcc6228995cdd94457a7f365125ac43ec968d579c55c30c98",
49
  "bytes": 874
50
  },
51
  "model.safetensors": {
52
+ "url": "https://huggingface.co/nirca/nirca-mini/resolve/main/checkpoints/4522274k/model.safetensors",
53
+ "sha256": "2da5f13ec471ba9cc6f6f2cbdd9eccb6c54d2667ae610141b917daaf829f2d4d",
54
  "bytes": 32638048
55
  }
56
  }
latest/config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "name": "nirca-mini",
3
- "step": 119196,
4
- "tokens": 4516803395,
5
  "quant": "ternary",
6
  "quant_scope": "all",
7
  "n_experts": 32,
 
1
  {
2
  "name": "nirca-mini",
3
+ "step": 119408,
4
+ "tokens": 4522274480,
5
  "quant": "ternary",
6
  "quant_scope": "all",
7
  "n_experts": 32,
latest/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d4b3bcc71ce2e9dd7442957e7c5304b8eb85aacbb64b183194e465add039e5d8
3
  size 32638048
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2da5f13ec471ba9cc6f6f2cbdd9eccb6c54d2667ae610141b917daaf829f2d4d
3
  size 32638048