bartowski commited on
Commit
0e695a2
1 Parent(s): 14a0225

Llamacpp quants

Browse files
Files changed (28) hide show
  1. .gitattributes +26 -0
  2. New-Dawn-Llama-3-70B-32K-v1.0-IQ1_M.gguf +3 -0
  3. New-Dawn-Llama-3-70B-32K-v1.0-IQ2_M.gguf +3 -0
  4. New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XS.gguf +3 -0
  5. New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XXS.gguf +3 -0
  6. New-Dawn-Llama-3-70B-32K-v1.0-IQ3_M.gguf +3 -0
  7. New-Dawn-Llama-3-70B-32K-v1.0-IQ3_XXS.gguf +3 -0
  8. New-Dawn-Llama-3-70B-32K-v1.0-IQ4_XS.gguf +3 -0
  9. New-Dawn-Llama-3-70B-32K-v1.0-Q2_K.gguf +3 -0
  10. New-Dawn-Llama-3-70B-32K-v1.0-Q2_K_L.gguf +3 -0
  11. New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_M.gguf +3 -0
  12. New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_S.gguf +3 -0
  13. New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_XL.gguf +3 -0
  14. New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_L.gguf +3 -0
  15. New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf +3 -0
  16. New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00001-of-00002.gguf +3 -0
  17. New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00002-of-00002.gguf +3 -0
  18. New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_M.gguf +3 -0
  19. New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00001-of-00002.gguf +3 -0
  20. New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00002-of-00002.gguf +3 -0
  21. New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00001-of-00002.gguf +3 -0
  22. New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00002-of-00002.gguf +3 -0
  23. New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00001-of-00004.gguf +3 -0
  24. New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00002-of-00004.gguf +3 -0
  25. New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00003-of-00004.gguf +3 -0
  26. New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00004-of-00004.gguf +3 -0
  27. New-Dawn-Llama-3-70B-32K-v1.0.imatrix +3 -0
  28. README.md +105 -0
.gitattributes CHANGED
@@ -33,3 +33,29 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ1_M.gguf filter=lfs diff=lfs merge=lfs -text
37
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text
38
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XS.gguf filter=lfs diff=lfs merge=lfs -text
39
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XXS.gguf filter=lfs diff=lfs merge=lfs -text
40
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ3_M.gguf filter=lfs diff=lfs merge=lfs -text
41
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text
42
+ New-Dawn-Llama-3-70B-32K-v1.0-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text
43
+ New-Dawn-Llama-3-70B-32K-v1.0-Q2_K.gguf filter=lfs diff=lfs merge=lfs -text
44
+ New-Dawn-Llama-3-70B-32K-v1.0-Q2_K_L.gguf filter=lfs diff=lfs merge=lfs -text
45
+ New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
46
+ New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
47
+ New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_XL.gguf filter=lfs diff=lfs merge=lfs -text
48
+ New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_L.gguf filter=lfs diff=lfs merge=lfs -text
49
+ New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
50
+ New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
51
+ New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
52
+ New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
53
+ New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
54
+ New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
55
+ New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00001-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
56
+ New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00002-of-00002.gguf filter=lfs diff=lfs merge=lfs -text
57
+ New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00001-of-00004.gguf filter=lfs diff=lfs merge=lfs -text
58
+ New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00002-of-00004.gguf filter=lfs diff=lfs merge=lfs -text
59
+ New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00003-of-00004.gguf filter=lfs diff=lfs merge=lfs -text
60
+ New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00004-of-00004.gguf filter=lfs diff=lfs merge=lfs -text
61
+ New-Dawn-Llama-3-70B-32K-v1.0.imatrix filter=lfs diff=lfs merge=lfs -text
New-Dawn-Llama-3-70B-32K-v1.0-IQ1_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:746467663be5cc4b6e915b3f6df0cc87cbbf6ea1729426a3cfeaa510273a4830
3
+ size 16751196000
New-Dawn-Llama-3-70B-32K-v1.0-IQ2_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2e257134656648cf2e3402eb66f11968f157a9a56dca61d88ffd191c6710610
3
+ size 24119293792
New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df79c3b48c13000829bb2359f0eb93b35b0d9aea36c1d3c089b9dc2efffa9b86
3
+ size 21142108000
New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XXS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3999d9f23a6f718ac909684decec7188acfc8974462b1e11d0cc4ead53d09f05
3
+ size 19097384800
New-Dawn-Llama-3-70B-32K-v1.0-IQ3_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78db2b65cf60392ff1b2adb14bfa8423951a4f0482a0c10aad2c4406a5a3de58
3
+ size 31937034080
New-Dawn-Llama-3-70B-32K-v1.0-IQ3_XXS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3c1d6b407f1f83a6251635e5a55c0b75c3d246cc81bb34d27f53707762f53d6
3
+ size 27469494112
New-Dawn-Llama-3-70B-32K-v1.0-IQ4_XS.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f4f73bb205f7ece0cb98655114a2d4857cb87eb915c5d789511ad1c5d66068c
3
+ size 37902661472
New-Dawn-Llama-3-70B-32K-v1.0-Q2_K.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31668bef7f3f8b46f7521fdce31ed9cadd21925cc2201a7ee18dca8ba1e27f7b
3
+ size 26375108448
New-Dawn-Llama-3-70B-32K-v1.0-Q2_K_L.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06fb82e4ac9e8b8a4b6c3edd786deab07064f012be61028c2c98e921adcfa33e
3
+ size 29371168608
New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad162294f0bbc2efa2166bbbd777312d2ae0702021098785845b48dcd819ef2f
3
+ size 34267494240
New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_S.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abbc84551a7d8f72a3c74bed8bc6cf0cd2aa9d7546cc81f31b06442155b216a3
3
+ size 30912051040
New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_XL.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:289ec1eccf2dbf0077d8980171348b9b017534fd7f8b0ce75a18382b7d215239
3
+ size 40029943648
New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_L.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c49eb33416f6350f25e043982b158f73c1919b5b0508c022492a5b7eb4b4b08
3
+ size 45270202208
New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30561ae5decac4ad46775c76a9a40fb43436ade96bc132b4b9cc6749b9e2f448
3
+ size 42520393568
New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21c6944a93fb00c253994158e227d7d42d2575e23a968f474938d1d6cdbea237
3
+ size 39922781152
New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:63637c64b81ae265bcfbdc153884d5d8dd798182a0e250a92d40b532be1fbdb4
3
+ size 12645510208
New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_M.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea149ec542da626609ee1c4cc1a7e37ec7c3731f4e50d21f39f6703cb8a34b00
3
+ size 49949816672
New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb15e0ae1f5b7a150a6bbaa929d8fe00238b1b4fb34bab0ca9eb122cf0180022
3
+ size 39829946720
New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4831070524e6a98a1f000a8b1fb07bb9bbf1057270f8ac10fc2b8d9a64d748f5
3
+ size 18058196672
New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00001-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e2abfeee7d3c33759db013bc35b37f6df8c62e4f7bcbab9c7a360c0d1d612e1
3
+ size 39766533664
New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0-00002-of-00002.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1f652d7a38d64e5a3dc3be2b48ccf27e23dd87ac60c274a9cf67b78e0b289bc
3
+ size 35208516096
New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00001-of-00004.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74d08e634428dd74498457b17655dff5d958971dfed5e7553f86cab578b696b4
3
+ size 39997700640
New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00002-of-00004.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14a14051df395972ed9353727c2e251f737789cf92487b17f3ab92af0c7b14c1
3
+ size 39981658528
New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00003-of-00004.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:862e7db3adc692040fec148f5411a9d009c722cefa2a5a08ac7f710af79bf39f
3
+ size 39830630656
New-Dawn-Llama-3-70B-32K-v1.0-f16.gguf/New-Dawn-Llama-3-70B-32K-v1.0-f16-00004-of-00004.gguf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0efa9d597b986293ab41de183df7860bc6cc1ba53d94afb9e5e1159c8944ab9
3
+ size 21307923264
New-Dawn-Llama-3-70B-32K-v1.0.imatrix ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2407d82fc9f1d88d171b09d4f36bfc7d29141d6f9cba5c314e5f6f19d721ec1
3
+ size 24922299
README.md ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - bosonai/Higgs-Llama-3-70B
4
+ - abacusai/Smaug-Llama-3-70B-Instruct-32K
5
+ - Sao10K/L3-70B-Euryale-v2.1
6
+ - abacusai/Smaug-Llama-3-70B-Instruct
7
+ - turboderp/Cat-Llama-3-70B-instruct
8
+ library_name: transformers
9
+ tags:
10
+ - mergekit
11
+ - merge
12
+ - Not-for-all-Audiences
13
+ license: llama3
14
+ quantized_by: bartowski
15
+ pipeline_tag: text-generation
16
+ ---
17
+
18
+ ## Llamacpp imatrix Quantizations of New-Dawn-Llama-3-70B-32K-v1.0
19
+
20
+ Using <a href="https://github.com/ggerganov/llama.cpp/">llama.cpp</a> release <a href="https://github.com/ggerganov/llama.cpp/releases/tag/b3197">b3197</a> for quantization.
21
+
22
+ Original model: https://huggingface.co/sophosympatheia/New-Dawn-Llama-3-70B-32K-v1.0
23
+
24
+ All quants made using imatrix option with dataset from [here](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8)
25
+
26
+ ## Prompt format
27
+
28
+ ```
29
+ <|begin_of_text|><|start_header_id|>system<|end_header_id|>
30
+
31
+ {system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
32
+
33
+ {prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
34
+
35
+
36
+ ```
37
+
38
+ ## Download a file (not the whole branch) from below:
39
+
40
+ | Filename | Quant type | File Size | Description |
41
+ | -------- | ---------- | --------- | ----------- |
42
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/tree/main/New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf) | Q8_0 | 74.97GB | Extremely high quality, generally unneeded but max available quant. |
43
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/tree/main/New-Dawn-Llama-3-70B-32K-v1.0-Q6_K.gguf) | Q6_K | 57.88GB | Very high quality, near perfect, *recommended*. |
44
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/tree/main/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_L.gguf) | Q5_K_L | 52.56GB | *Experimental*, uses f16 for embed and output weights. Please provide any feedback of differences. High quality, *recommended*. |
45
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q5_K_M.gguf) | Q5_K_M | 49.94GB | High quality, *recommended*. |
46
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_L.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_L.gguf) | Q4_K_L | 45.27GB | *Experimental*, uses f16 for embed and output weights. Please provide any feedback of differences. Good quality, uses about 4.83 bits per weight, *recommended*. |
47
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf) | Q4_K_M | 42.52GB | Good quality, uses about 4.83 bits per weight, *recommended*. |
48
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ4_XS.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ4_XS.gguf) | IQ4_XS | 37.90GB | Decent quality, smaller than Q4_K_S with similar performance, *recommended*. |
49
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_M.gguf) | Q3_K_M | 34.26GB | Even lower quality. |
50
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ3_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ3_M.gguf) | IQ3_M | 31.93GB | Medium-low quality, new method with decent performance comparable to Q3_K_M. |
51
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_S.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q3_K_S.gguf) | Q3_K_S | 30.91GB | Low quality, not recommended. |
52
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ3_XXS.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ3_XXS.gguf) | IQ3_XXS | 27.46GB | Lower quality, new method with decent performance, comparable to Q3 quants. |
53
+ | [New-Dawn-Llama-3-70B-32K-v1.0-Q2_K.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-Q2_K.gguf) | Q2_K | 26.37GB | Very low quality but surprisingly usable. |
54
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ2_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ2_M.gguf) | IQ2_M | 24.11GB | Very low quality, uses SOTA techniques to also be surprisingly usable. |
55
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XS.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XS.gguf) | IQ2_XS | 21.14GB | Lower quality, uses SOTA techniques to be usable. |
56
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XXS.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ2_XXS.gguf) | IQ2_XXS | 19.09GB | Lower quality, uses SOTA techniques to be usable. |
57
+ | [New-Dawn-Llama-3-70B-32K-v1.0-IQ1_M.gguf](https://huggingface.co/bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF/blob/main/New-Dawn-Llama-3-70B-32K-v1.0-IQ1_M.gguf) | IQ1_M | 16.75GB | Extremely low quality, *not* recommended. |
58
+
59
+ ## Downloading using huggingface-cli
60
+
61
+ First, make sure you have hugginface-cli installed:
62
+
63
+ ```
64
+ pip install -U "huggingface_hub[cli]"
65
+ ```
66
+
67
+ Then, you can target the specific file you want:
68
+
69
+ ```
70
+ huggingface-cli download bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF --include "New-Dawn-Llama-3-70B-32K-v1.0-Q4_K_M.gguf" --local-dir ./
71
+ ```
72
+
73
+ If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:
74
+
75
+ ```
76
+ huggingface-cli download bartowski/New-Dawn-Llama-3-70B-32K-v1.0-GGUF --include "New-Dawn-Llama-3-70B-32K-v1.0-Q8_0.gguf/*" --local-dir New-Dawn-Llama-3-70B-32K-v1.0-Q8_0
77
+ ```
78
+
79
+ You can either specify a new local-dir (New-Dawn-Llama-3-70B-32K-v1.0-Q8_0) or download them all in place (./)
80
+
81
+ ## Which file should I choose?
82
+
83
+ A great write up with charts showing various performances is provided by Artefact2 [here](https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9)
84
+
85
+ The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.
86
+
87
+ If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.
88
+
89
+ If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.
90
+
91
+ Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.
92
+
93
+ If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.
94
+
95
+ If you want to get more into the weeds, you can check out this extremely useful feature chart:
96
+
97
+ [llama.cpp feature matrix](https://github.com/ggerganov/llama.cpp/wiki/Feature-matrix)
98
+
99
+ But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.
100
+
101
+ These I-quants can also be used on CPU and Apple Metal, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.
102
+
103
+ The I-quants are *not* compatible with Vulcan, which is also AMD, so if you have an AMD card double check if you're using the rocBLAS build or the Vulcan build. At the time of writing this, LM Studio has a preview with ROCm support, and other inference engines have specific builds for ROCm.
104
+
105
+ Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski